CN111753692A - Target object extraction method, product detection method, device, computer and medium - Google Patents

Abstract

The invention provides a target object extraction method, a product detection method, a device, a computer and a medium. The target object extraction method includes acquiring an image; inputting the image into a pre-trained training model for feature detection to obtain the feature information of the image; According to the feature information, the area of interest of the image is obtained from the image; according to the area of interest, the coverage area of the target object in the image is obtained; based on the coverage area of the target object in the image, the target contour of the target object in the image is obtained; Object contour, extracting the target object from the image. By training the model to perform feature detection on the image, the feature information of the image is obtained, the region of interest in the image is determined, the position of the target object in the image is obtained, the contour of the target object in the image is obtained, and the target object is extracted from the image. The automatic extraction of the target object in the image is realized, which makes the extraction of the refrigerator panel more accurate and improves the efficiency.

Description

Target object extraction method, product detection method, device, computer and medium

technical field

The invention relates to the technical field of image recognition, and in particular, to a target object extraction method, a product detection method, a device, a computer and a medium.

Background technique

In the field of modern manufacturing production, image registration is a key link in product quality inspection based on computer vision, and the effect of image registration directly affects the speed and effect of quality inspection. The configuration effect depends to a large extent on the selection of feature regions, so it is particularly important to select an ideal feature region. At present, in the quality inspection of refrigerator panels, most of the feature areas are selected manually or semi-automatically. Manual selection is time-consuming, labor-intensive and inefficient. At the same time, the manual selection of feature areas varies from person to person, which is not objective enough, which affects the efficiency and efficiency of the existing detection system. Detection accuracy.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide a target object extraction method, a product detection method, an apparatus, a computer and a medium for the above technical problems.

An image target object extraction method, comprising:

get image;

Inputting the image into a pre-trained training model for feature detection to obtain feature information of the image;

Selecting a preset number of regions of interest for each of the feature information;

Perform binary classification of the foreground and background on the image including the region of interest, obtain pixel values of the foreground and background, classify based on a preset pixel threshold, and obtain the coverage of the target object in the image from the region of interest area;

Obtain the target contour of the target object in the image based on the coverage area of the target object in the image;

The target object is extracted from the image according to the target contour in the image.

In one embodiment, before the step of acquiring an image, it further includes:

Get training images;

Carry out contour calibration on the target object of the training image, and generate a training sample, and the training sample is an image file demarcating the contour of the target object;

The training samples are input into a convolutional neural network for learning, and the training model including the feature information of the target object of each training image is obtained.

In one embodiment, the step of selecting a preset number of regions of interest for each of the feature information includes:

For each of the feature information, select a preset number of candidate regions of interest;

Binary classification of foreground and background is performed on the candidate regions of interest, and a region of interest of the image is selected from a preset number of the candidate regions of interest.

In one embodiment, the step of obtaining the target contour of the target object in the image based on the coverage area of the target object in the image includes:

The target contour of the target object is extracted from the image based on the coverage area of the target object in the image, using a target detection method and a threshold segmentation method.

In one embodiment, the step of extracting the target object from the image according to the target contour in the image includes:

performing edge processing on the extracted target contour to obtain edge position information of the target contour;

The target object is extracted from the image based on edge position information of the target contour.

In one embodiment, the step of performing edge processing on the extracted target contour to obtain edge position information of the target contour includes:

Based on the target contour, obtain a mask image of the image including black and white;

The Canny detection algorithm is used to perform edge processing on the black and white mask image to obtain edge position information of the target contour.

In one embodiment, the step of extracting the target object from the image based on the edge position information of the target contour includes:

Using a threshold segmentation method, the gray value of each pixel in the image is obtained;

Compare the grayscale value of each pixel in the image with a preset grayscale threshold, and perform binarization on the comparison result of the grayscale value of each pixel in the image and the preset grayscale threshold to obtain two The image after value processing;

The target object is extracted from the binarized image based on the edge position information of the target contour.

A product testing method, comprising:

According to the image target object extraction method described in any of the above embodiments, the target object in the product image is extracted, and the image information of the target object is obtained, and whether the product meets the preset requirements is judged according to the image information of the target object.

An image target object extraction device, comprising:

Image acquisition module for acquiring images;

a feature information obtaining module, used for inputting the image into a pre-trained training model for feature detection to obtain feature information of the image;

a region of interest obtaining module, used for selecting a preset number of regions of interest for each of the feature information;

The target position information obtaining module is used to perform binary classification of the foreground and background on the image including the region of interest, obtain the pixel value of the foreground and background, classify based on a preset pixel threshold, and obtain from the region of interest The target object is in the coverage area of the image;

a target contour obtaining module, configured to obtain the target contour of the target object in the image based on the coverage area of the target object in the image;

A target object extraction module, configured to extract the target object from the image according to the target contour in the image.

On the target object extraction method, product detection method, device, computer and medium, the pre-trained training model is used to perform feature detection on the image based on semantic segmentation technology, and the feature information of the image is obtained, and then the region of interest in the image is determined to locate. The position of the target object in the image is obtained, and based on the position in the image, the contour of the target object in the image is obtained, and the target object is extracted from the image according to the contour. The automatic extraction of the target object in the image is realized without manual selection, which makes the extraction of the refrigerator panel more accurate, faster and more convenient, effectively saves time, streamlines manpower, and improves efficiency.

Description of drawings

1A is a schematic flowchart of an image target object extraction method in one embodiment;

1B is a schematic flowchart of an image target object extraction method in another embodiment;

1C is a schematic flowchart of an image target object extraction method in another embodiment;

1D is a schematic flowchart of an image target object extraction method in yet another embodiment;

Fig. 2 is a structural block diagram of an image target object extraction apparatus in one embodiment;

Fig. 3 is the internal structure diagram of the computer device in one embodiment;

4 is a schematic flowchart of a training process of a training model in one embodiment;

FIG. 5 is a schematic flowchart of an image target object extraction process in one embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

It should be understood that the image target object extraction method in this application can be applied to the extraction of the panels of refrigerators, and can also be applied to the extraction of panels in other manufacturing products, or in other scenarios where a portrait of an object needs to be extracted from the image. , each scene in this application will not be repeated one by one. In the following embodiments, the method for extracting an image target object is applied to the extraction of a refrigerator panel for further explanation. It should be understood that the target object extracted in this application refers to the image or figure of the target object in the image.

In one embodiment, as shown in FIG. 1A, a method for extracting an image target object is provided,

Step 110, acquire an image.

Specifically, the image is an image of the target object to be extracted. For example, the image is an image of a refrigerator, and the image of the refrigerator includes a refrigerator panel. Therefore, the target object is a refrigerator panel, and the target object is an image of a refrigerator panel, which may be a front panel, a rear panel, or a side panel.

Step 120: Input the image into a pre-trained training model for feature detection to obtain feature information of the image.

Specifically, the training model is pre-trained and can perform feature detection on the features of the image. In this embodiment, the training model adopts the branch network of the segmentation deep learning Mask-RCNN framework to perform feature detection on the image, and the training model includes a pre-trained feature information set of the target object. In this embodiment, the training model performs feature detection on the image based on semantic segmentation, and then obtains the feature information of the image. In this way, the feature information of the input image can be detected by inputting the image into the training model.

In one embodiment, step 120 includes preprocessing the image to obtain a preprocessed image; inputting the preprocessed image into a pretrained training model for learning to obtain features of the image information.

In this embodiment, irrelevant information in the image is eliminated through the preprocessing of the image, useful real information is retained, the image is denoised, the detectability of the relevant information is enhanced, and the data is simplified. It should be understood that when processing a color image, three channels need to be processed in sequence, and the time overhead will be very large. The section stores the gray value, and the gray scale range is 0-255, which reduces the data amount of the pixels of the image, which is beneficial to improve the overall processing speed, and needs to reduce the amount of data to be processed.

Step 130: Select a preset number of regions of interest for each of the feature information.

In this step, for each feature information in the image, at least one region of interest is selected respectively. The region of interest is an image region selected from the image. The selection method may be selection according to the quantity of the feature information, or may be selected according to the position of the feature information, and the selection method may also be manual selection. It should be understood that the selection of the region of interest for the feature information may be implemented by using the prior art, and redundant description is not described in this embodiment.

Step 140: Perform binary classification of the foreground and background on the image including the region of interest, obtain pixel values of the foreground and background, perform classification based on a preset pixel threshold, and obtain the target object in the region of interest from the region of interest. The coverage area of the image.

Specifically, after binary classification of foreground and background is performed on the image, the pixel values of the foreground and background of the image are obtained, a preset pixel threshold is obtained, and each region of interest is classified based on the pixel threshold. The comparison between the pixel value and the pixel value of the background and the preset pixel threshold value, because the foreground and the background are binarized, so that the pixels of the foreground and the background are converted into grayscale values, and then based on the grayscale value and the preset pixel threshold value In contrast, the target object can be divided into the target object that is larger than the preset pixel threshold, and the background can be divided into the background smaller than the preset pixel threshold, so as to determine the coverage area of the target object, so as to select the target object from the region of interest in the image. coverage area. In this step, the region of interest is classified, and the position region of the target object in the image is located and searched, that is, the position information of the target object in the image is obtained. Specifically, according to the pixel value difference between the foreground and the background of the image, according to the preset pixel threshold, the region of interest is classified, and the region of interest corresponding to the target object is obtained, so as to locate the position region of the target object in the image, Then, the coverage area of the target object in the image is obtained.

Step 150: Obtain a target contour of the target object in the image based on the coverage area of the target object in the image.

Specifically, the target contour is the contour of the target object in the image. In this embodiment, after the feature detection of semantic segmentation is performed on the image, the target detection method and the threshold segmentation method are used to extract the contour of the target object in the image, Then the target contour of the target object is obtained.

Step 160: Extract the target object from the image according to the target contour in the image.

Specifically, after determining the target contour of the target object in the image, according to the target contour, and then according to the threshold segmentation method, the target object in the target contour is extracted from the image, so as to automatically extract the target object in the image, It should be understood that the extracted target object is an image of the target object. For example, extract the target object and obtain the image information of the target object. Therefore, the image of the panel of the refrigerator is automatically extracted from the image of the refrigerator without manual selection, which makes the extraction of the refrigerator panel more accurate, faster and more convenient, effectively saves time, streamlines manpower, and improves efficiency.

In the above embodiment, the pre-trained training model is used to perform feature detection on the image based on the semantic segmentation technology to obtain the feature information of the image, and then determine the region of interest in the image, so as to locate and obtain the position of the target object in the image, and based on position in the image, obtain the contour of the target object in the image, and extract the target object from the image according to the contour. The automatic extraction of the target object in the image is realized without manual selection, which makes the extraction of the refrigerator panel more accurate, faster and more convenient, effectively saves time, streamlines manpower, and improves efficiency.

In one embodiment, as shown in FIG. 1B , before step 110, it further includes:

Step 101, acquiring training images.

Step 102: Perform contour calibration on the target object of the training image to generate a training sample, where the training sample is an image file demarcating the contour of the target object.

Step 103: Input the training samples into a convolutional neural network for learning, and obtain the training model including the feature information of the target object of each of the training images.

Specifically, the training image is an image of the same type as the image in the foregoing embodiment, and the training image includes the target object. For example, the training image is an image of a refrigerator, including images of refrigerators of different models, while the images in the foregoing embodiment It is also an image of a refrigerator, and a panel is included in the image of each refrigerator. Before the training image is input into the convolutional neural network for learning, the training image is qualified to generate an image file that demarcates the outline of the target object in the training image, and the image file is the training sample. The training sample is input into the convolutional neural network for learning, and a training model is obtained. The training model contains a set of feature information of the target object of the training image. In this way, through the training of a large number of training images, different refrigerator panels can be obtained. A collection of feature information.

In one embodiment, various types of refrigerator panel images are collected, and the LabelMe tool is used to perform detailed outline calibration of the target object in the training image. It is worth mentioning that the LabelMe tool can be used to create customized labeling tasks or execute images. callout. After the outline annotation of the target object is completed, a JSON (JavaScript ObjectNotation,, JS Object Notation) file is generated. The JSON file is the training sample. Subsequently, each JSON file is converted into a DataSet, and the DataSet contains: img.png, info.yaml, label.png, label_names.txt, label_viz.png, and generates a mask dataset. The convolutional neural network reads the mask data set based on the mask-rcnn algorithm for model training. After the training is completed, the feature information set of the refrigerator panel is generated. The feature information set can also be called the feature data set. characteristic information.

In one embodiment, as shown in Figure 1C, step 130 includes:

Step 131: For each of the feature information, select a preset number of candidate regions of interest.

Step 132 , perform binary classification of the foreground and background on the candidate regions of interest, and select the regions of interest of the image from a preset number of the candidate regions of interest.

In this embodiment, a preset number of candidate regions of interest are respectively set for each feature information. Specifically, a preset number of candidate regions of interest (ROI, Region Of Interest) are respectively set for each point of the obtained sets of multiple feature information, and the preset number is preset. These candidate regions of interest are sent to the regional network for binary classification of the foreground and background, and some candidate regions of interest are filtered out to screen out suitable regions of interest. It should be understood that the area network is a candidate for performing an area frame on the input image, and uses the edge, texture, color, color change and other information of the image to select areas in the image that may contain objects. In this embodiment, the candidate region of interest is sent to the regional network for binary classification of the foreground and background, and the candidate is interested in the candidate according to the pixel value difference between the foreground and the background of the image according to the preset pixel threshold. Regions are classified to filter out regions of interest.

In one embodiment, the step of obtaining the target contour of the target object in the image based on the coverage area of the target object in the image includes: In the coverage area, the target contour of the target object is extracted from the image by using the target detection method and the threshold segmentation method.

Specifically, in this embodiment, the position information of the target object is detected through semantic segmentation in the previous embodiment, and based on the position information, the target detection method and the threshold segmentation method are used to separate the target object in the image in the complex background. The contour is extracted to obtain the target contour. It is worth mentioning that the threshold segmentation algorithm can use fast rcnn, yolo, SSD and other algorithms. Through the target detection method and the threshold segmentation method, the target contour can be extracted efficiently and accurately.

In one embodiment, as shown in Figure ID, step 160 includes:

Step 161: Perform edge processing on the extracted target contour to obtain edge position information of the target contour.

Step 162: Extract the target object from the image based on the edge position information of the target contour.

In this embodiment, through edge processing, an optimal edge of the target contour is found, and accurate position information of the edge of the target contour is obtained, and then based on the edge position information, the position and range of the target contour can be accurately located, thereby improving the accuracy of the target contour. The accuracy of the target object extracted in the image.

In one embodiment, the step of performing edge processing on the extracted target contour to obtain edge position information of the target contour includes: based on the target contour, obtaining a mask including black and white for the image image; using the Canny detection algorithm to perform edge processing on the black and white mask image to obtain the edge position information of the target contour.

In this embodiment, after the target contour in the image is extracted, a mask image containing only black and white colors is extracted, and the image Canny detection algorithm is used to perform edge processing on the mask image to obtain an optimal edge. In this way, The edge of the target contour can be made more accurate, which is beneficial to improve the extraction accuracy of the target object.

In one embodiment, the step of extracting the target object from the image based on the edge position information of the target contour includes: using a threshold segmentation method to obtain the gray value of each pixel in the image; Compare the grayscale value of each pixel in the image with a preset grayscale threshold, and perform binarization on the comparison result of the grayscale value of each pixel in the image and the preset grayscale threshold to obtain two The binarized image; extracting the target object from the binarized image based on the edge position information of the target contour.

Specifically, the threshold segmentation method is used to determine a certain gray value of each pixel in the image that is in the gray scale range, and the gray value of each pixel in the obtained image is compared with that obtained by the threshold segmentation method in the previous step. The gray threshold is compared, and the appropriate threshold is selected for segmentation, and the binarization algorithm is processed, and finally the target object is extracted. The appropriate threshold can be set according to the pixel grayscale of the target object and the background, and the preset grayscale threshold can be obtained by setting the threshold at the same time. Through the above process, the influence of illumination caused by the environment on the features of the image can be reduced.

It should be understood that, in the threshold separation method, an appropriate pixel value is selected as the preset grayscale threshold, and the preset grayscale threshold can be manually set according to the pixel value of the analyzed image. The image is processed into a high-contrast, easy-to-recognize image using the preset grayscale threshold as a limit. For example, if it is greater than the preset grayscale threshold, set it as white, and if it is less than or equal to the preset grayscale threshold, set it as black, the output black and white image, the white is the target object, and the black is the background other than the target object in the image. In this way, the target object can be separated from the image to realize the extraction of the target object.

The following is a specific example:

First, the refrigerator panel graphic marking training. Please refer to Figure 4 and use the deep learning method of artificial intelligence to collect data on the feature images of the refrigerator panels, and then mark the features of each type of refrigerator panels, and use the convolutional neural network algorithm to train the model according to the sample information to obtain semantic recognition. panel object.

Second, feature detection search. Feature detection is performed on candidate regions through a branched network of an instance segmentation deep learning Mask-RCNN framework. Please refer to Figure 5, first preprocess the input image, and then input it into a pre-trained neural network to obtain the corresponding data information; then, set a predetermined value for each point in the image data information A number of ROIs are obtained to obtain multiple candidate ROIs; these candidate ROIs are sent to the regional network for binary classification of the foreground and background, and some candidate ROIs are filtered out; finally, these ROIs are classified, and the location area of the refrigerator panel is located and searched.

Finally, target object extraction. The target features are detected by semantic segmentation, and the general outline of the panel is extracted from the complex background by using target detection and threshold segmentation technology, and a mask containing only black and white colors is extracted. The mask is processed by the image canny detection algorithm for edge processing. , find an optimal edge, identify the actual edge in the image as much as possible, and then use the threshold segmentation method to determine a certain gray value of each pixel in the gray scale range of the refrigerator panel image. The gray value of each pixel in the image is compared with the previously determined threshold, and the segmentation process is performed, and the binarization algorithm is processed. Selecting the appropriate threshold can reduce the influence of the illumination caused by the environment on the image features, and finally extract the features automatically. Refrigerator panel image.

In this embodiment, the threshold segmentation algorithm may also adopt algorithms such as fast rcnn, yolo, and SSD.

In this application, the artificial intelligence deep learning algorithm, semantic segmentation technology, and image edge detection are used to detect the area of the refrigerator panel area, perform pixel-level classification, automatically, quickly and accurately extract refrigerator panel information, and improve the work efficiency of quality inspectors.

The present application uses artificial intelligence image segmentation algorithms to automatically extract features for refrigerator panels in the manufacturing industry. Compared with traditional manual or semi-automatic methods, the method proposed in this application can achieve objectification, standardization and unification of automatic feature extraction based on deep learning algorithm and by training refrigerator panel information data, and the accuracy is higher than that of manual selection of feature regions. There is a great improvement, which solves the problem of artificial selection of feature areas that vary from person to person and is not objective enough, and improves the efficiency and detection accuracy of the detection system.

This method proposes an automatic extraction method of refrigerator panel features based on artificial intelligence. Using artificial intelligence deep learning algorithm and semantic segmentation technology, the automatic and accurate extraction of refrigerator panels in manufacturing industry is realized. This method solves the problem of manual calibration or semi-automatic calibration in the traditional sense. The problem of taking pictures in a specific environment can be automatically and accurately extracted, fast and convenient, and effectively save time and manpower.

In one embodiment, a product detection method is characterized in that it includes extracting a target object in a product image according to the image target object extraction method described in any of the above embodiments, and obtaining image information of the target object, according to The image information of the target object determines whether the product meets the preset requirements.

In this embodiment, the product detection method may also be referred to as a defect detection method of a target object of a product image, and the method is used to detect a defect in a certain part of the product, for example, the method is used to detect a defect of a panel of a refrigerator. Extract the target object of the product through the target object extraction method. For example, first extract the refrigerator panel from the refrigerator image, so as to accurately locate the refrigerator panel, and then judge whether the product meets the preset requirements according to the image information of the target object, and then analyze the refrigerator panel. Perform defect detection. In addition to the steps of the method for extracting the target object of the image, the product detection method also includes the step of detecting the defect of the target object. It is worth mentioning that the step of detecting the defect of the target object can be realized by using the existing technology. It can also be implemented by using a conventional defect detection method in the art. These defect detection methods are known to those skilled in the art, and are not repeated in this embodiment.

In one embodiment, as shown in FIG. 2, an image target object extraction apparatus is provided, including:

an image acquisition module 210, configured to acquire an image;

A feature information obtaining module 220, configured to input the image into a pre-trained training model for feature detection to obtain feature information of the image;

a region of interest obtaining module 230, configured to select a preset number of regions of interest for each of the feature information;

The target position information obtaining module 240 is configured to perform binary classification of the foreground and background on the image including the region of interest, obtain pixel values of the foreground and background, classify based on a preset pixel threshold, and extract the image from the region of interest. Obtain the coverage area of the target object in the image;

A target contour obtaining module 250, configured to obtain the target contour of the target object in the image based on the coverage area of the target object in the image;

The target object extraction module 260 is configured to extract the target object from the image according to the target contour in the image.

In one embodiment, the image target object extraction apparatus further includes:

A training image acquisition module for acquiring training images;

A training image marking module is used to demarcate the outline of the target object of the training image, and generate a training sample, and the training sample is an image file that has demarcated the outline of the target object;

A training model generation module is configured to input the training samples into a convolutional neural network for learning, and obtain the training model including the feature information of the target object of each of the training images.

In one embodiment, the region of interest obtaining module includes:

a candidate region of interest obtaining unit for selecting a preset number of candidate regions of interest for each of the feature information;

A region of interest screening unit, configured to perform binary classification of the foreground and background on the candidate region of interest, and select the region of interest of the image from a preset number of the candidate regions of interest.

In one embodiment, the target contour obtaining module is further configured to extract the target contour of the target object in the image by adopting a target detection method and a threshold segmentation method based on the coverage area of the target object in the image. .

In one embodiment, the target object extraction module includes:

an edge position information obtaining unit, configured to perform edge processing on the extracted target contour to obtain edge position information of the target contour;

A target object extraction unit, configured to extract the target object from the image based on edge position information of the target contour.

In one embodiment, the edge location information obtaining unit includes:

a mask image obtaining subunit, configured to obtain a mask image including black and white of the image based on the target contour;

The edge position information obtaining sub-unit is used for using the Canny detection algorithm to perform edge processing on the black and white mask image to obtain the edge position information of the target contour.

In one embodiment, the target object extraction unit includes:

a gray value obtaining subunit, used for obtaining the gray value of each pixel in the image by adopting a threshold segmentation method;

The binarization processing subunit is used to compare the gray value of each pixel in the image with a preset gray threshold, and compare the gray value of each pixel in the image with the preset gray threshold The result is binarized to obtain a binarized image;

The target object extraction subunit is used for extracting the target object from the binarized image based on the edge position information of the target contour.

For the specific limitation of the image target object extraction apparatus, reference may be made to the above limitation on the image target object extraction method, which will not be repeated here. Each module in the above-mentioned image target object extraction apparatus may be implemented in whole or in part by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided. Its internal structure diagram can be shown in Figure 3. The computer equipment includes a processor, memory, a network interface, a display screen, and an input device connected by a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium, an internal memory. The nonvolatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used for external computers to communicate over the network connection. The computer program, when executed by the processor, implements an image target object extraction method. The display screen of the computer equipment may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment may be a touch layer covered on the display screen, or a button, a trackball or a touchpad set on the shell of the computer equipment , or an external keyboard, trackpad, or mouse.

Those skilled in the art can understand that the structure shown in FIG. 3 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied. Include more or fewer components than shown in the figures, or combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

get image;

Inputting the image into a pre-trained training model for feature detection to obtain feature information of the image;

Selecting a preset number of regions of interest for each of the feature information;

Perform binary classification of the foreground and background on the image including the region of interest, obtain pixel values of the foreground and background, classify based on a preset pixel threshold, and obtain the coverage of the target object in the image from the region of interest area;

Obtain the target contour of the target object in the image based on the coverage area of the target object in the image;

The target object is extracted from the image according to the target contour in the image.

In one embodiment, the processor further implements the following steps when executing the computer program:

Get training images;

Carry out contour calibration on the target object of the training image, and generate a training sample, and the training sample is an image file demarcating the contour of the target object;

The training samples are input into a convolutional neural network for learning, and the training model including the feature information of the target object of each training image is obtained.

In one embodiment, the processor further implements the following steps when executing the computer program:

For each of the feature information, select a preset number of candidate regions of interest;

Binary classification of foreground and background is performed on the candidate regions of interest, and a region of interest of the image is selected from a preset number of the candidate regions of interest.

In one embodiment, the processor further implements the following steps when executing the computer program:

The target contour of the target object is extracted from the image based on the coverage area of the target object in the image, using a target detection method and a threshold segmentation method.

In one embodiment, the processor further implements the following steps when executing the computer program:

performing edge processing on the extracted target contour to obtain edge position information of the target contour;

The target object is extracted from the image based on edge position information of the target contour.

In one embodiment, the processor further implements the following steps when executing the computer program:

Based on the target contour, obtain a mask image of the image including black and white;

The Canny detection algorithm is used to perform edge processing on the black and white mask image to obtain edge position information of the target contour.

In one embodiment, the processor further implements the following steps when executing the computer program:

Using a threshold segmentation method, the gray value of each pixel in the image is obtained;

Compare the grayscale value of each pixel in the image with a preset grayscale threshold, and perform binarization on the comparison result of the grayscale value of each pixel in the image and the preset grayscale threshold to obtain two The image after value processing;

The target object is extracted from the binarized image based on the edge position information of the target contour.

In one embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements a product detection method when the computer program is executed.

In one embodiment, a computer-readable storage medium is provided on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

get image;

Inputting the image into a pre-trained training model for feature detection to obtain feature information of the image;

Selecting a preset number of regions of interest for each of the feature information;

Perform binary classification of the foreground and background on the image including the region of interest, obtain pixel values of the foreground and background, classify based on a preset pixel threshold, and obtain the coverage of the target object in the image from the region of interest area;

Obtain the target contour of the target object in the image based on the coverage area of the target object in the image;

The target object is extracted from the image according to the target contour in the image.

In one embodiment, the computer program further implements the following steps when executed by the processor:

Get training images;

Carry out contour calibration on the target object of the training image, and generate a training sample, and the training sample is an image file demarcating the contour of the target object;

The training samples are input into a convolutional neural network for learning, and the training model including the feature information of the target object of each training image is obtained.

In one embodiment, the computer program further implements the following steps when executed by the processor:

For each of the feature information, select a preset number of candidate regions of interest;

Binary classification of foreground and background is performed on the candidate regions of interest, and a region of interest of the image is selected from a preset number of the candidate regions of interest.

In one embodiment, the computer program further implements the following steps when executed by the processor:

The target contour of the target object is extracted from the image based on the coverage area of the target object in the image, using a target detection method and a threshold segmentation method.

In one embodiment, the computer program further implements the following steps when executed by the processor:

performing edge processing on the extracted target contour to obtain edge position information of the target contour;

The target object is extracted from the image based on edge position information of the target contour.

In one embodiment, the computer program further implements the following steps when executed by the processor:

Based on the target contour, obtain a mask image of the image including black and white;

The Canny detection algorithm is used to perform edge processing on the black and white mask image to obtain edge position information of the target contour.

In one embodiment, the computer program further implements the following steps when executed by the processor:

Using a threshold segmentation method, the gray value of each pixel in the image is obtained;

Compare the grayscale value of each pixel in the image with a preset grayscale threshold, and perform binarization on the comparison result of the grayscale value of each pixel in the image and the preset grayscale threshold to obtain two The image after value processing;

The target object is extracted from the binarized image based on the edge position information of the target contour.

In one embodiment, a computer-readable storage medium is provided on which a computer program is stored, and when the computer program is executed by a processor, implements a product detection method.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other medium used in the various embodiments provided in this application may include non-volatile and/or volatile memory. Nonvolatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Road (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features It is considered to be the range described in this specification.

The above-mentioned embodiments only represent several embodiments of the present application, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the scope of protection of the patent of the present application shall be subject to the appended claims.

Claims (11)

1. An image target object extraction method, comprising:

acquiring an image;

inputting the image into a training model trained in advance for feature detection to obtain feature information of the image;

selecting a preset number of interested areas for each feature information;

performing foreground and background binary classification on the image containing the region of interest to obtain a foreground and background pixel value, performing classification based on a preset pixel threshold value, and obtaining a coverage area of a target object in the image from the region of interest;

obtaining a target contour of the target object in the image based on the coverage area of the target object in the image;

and extracting the target object from the image according to the target contour in the image.

2. The method of claim 1, further comprising, prior to the step of acquiring an image:

acquiring a training image;

carrying out contour calibration on a target object of the training image to generate a training sample, wherein the training sample is an image file with the contour of the target object calibrated;

and inputting the training samples into a convolutional neural network for learning to obtain the training model containing the characteristic information of the target object of each training image.

3. The method of claim 1, wherein the step of selecting a predetermined number of regions of interest for each of the feature information comprises:

selecting a preset number of candidate interesting regions for each characteristic information;

and carrying out foreground and background binary classification on the candidate interesting regions, and screening the interesting regions of the image from a preset number of the candidate interesting regions.

4. The method of claim 1, wherein the step of obtaining a target contour of the target object in the image based on a coverage area of the target object in the image comprises:

and extracting a target contour of the target object in the image by adopting a target detection method and a threshold segmentation method based on the coverage area of the target object in the image.

5. The method according to any one of claims 1-4, wherein said step of extracting said target object from said image based on a target contour in said image comprises:

performing edge processing on the extracted target contour to obtain edge position information of the target contour;

and extracting the target object from the image based on the edge position information of the target contour.

6. The method according to claim 5, wherein the step of performing edge processing on the extracted target contour to obtain edge position information of the target contour comprises:

obtaining a mask image containing black and white two colors of the image based on the target contour;

and performing edge processing on the black-white two-color mask image by adopting a Canny detection algorithm to obtain edge position information of the target contour.

7. The method of claim 5, wherein the step of extracting the target object from the image based on the edge position information of the target contour comprises:

acquiring the gray value of each pixel point in the image by adopting a threshold segmentation method;

comparing the gray value of each pixel point in the image with a preset gray threshold value, and performing binarization processing on the comparison result of the gray value of each pixel point in the image and the preset gray threshold value to obtain an image after binarization processing;

and extracting the target object from the image after the binarization processing based on the edge position information of the target contour.

8. A method of product inspection, comprising:

the image target object extraction method according to any one of claims 1 to 7, extracting a target object in a product image, obtaining image information of the target object, and judging whether the product meets a preset requirement according to the image information of the target object.

9. An image target object extraction device characterized by comprising:

the image acquisition module is used for acquiring an image;

the characteristic information acquisition module is used for inputting the image into a pre-trained training model for characteristic detection to acquire characteristic information of the image;

the interesting region obtaining module is used for selecting a preset number of interesting regions for each piece of characteristic information;

the target position information obtaining module is used for carrying out foreground and background binary classification on the image containing the region of interest to obtain a foreground and background pixel value, carrying out classification based on a preset pixel threshold value and obtaining a coverage area of a target object in the image from the region of interest;

a target contour obtaining module, configured to obtain a target contour of the target object in the image based on a coverage area of the target object in the image;

and the target object extraction module is used for extracting the target object from the image according to the target contour in the image.

10. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 8 when executing the computer program.

11. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.

Images