CN103533242A - Method and system for extracting and tracking cursor point in out-of-focus video - Google Patents

Abstract

The invention relates to a method for extracting and tracking a cursor point in an out-of-focus video, comprising the following steps: a. Importing the first frame image of the out-of-focus video; b. Delimiting one or more in the first frame image The image area is used as the search area of the cursor point, and each search area is used as a sub-image of the overall image to extract the cursor point and obtain the center coordinates of the cursor point; c. import the next frame image of the out-of-focus video; d. in the next frame Segment sub-images in the image; e. extract the cursor point in each sub-image, and obtain the center coordinates of the cursor point; f. correct the above-mentioned obtained cursor point center coordinates, and output the corrected cursor point center coordinates g. Update the search area; h. Repeat step c to step g until the last frame of image. The invention also relates to a system for extracting and tracking a cursor point in an out-of-focus video. The invention can reduce the false recognition rate and missed recognition rate, and improve the accuracy of data.

Description

Method and system for extracting and tracking cursor point in out-of-focus video

technical field

The invention relates to a method and system for extracting and tracking a cursor point in an out-of-focus video.

Background technique

Computer vision is an important branch of computer science and artificial intelligence. The ultimate goal of research is to enable computers to observe and understand things through acquired images like humans. Human motion capture is a fast-growing and highly concerned frontier in the field of computer vision. It is an important technology combining computer vision and biomechanics. It is widely used in film and television special effects, animation production, game production, virtual reality, and sports training. , biomechanical analysis and other fields.

Human body motion capture is the process of recovering human body posture parameters from image sequences captured by one or more cameras (the posture here refers to the movement of the torso, limbs and head, excluding small-scale movements such as facial expressions). Paste the cursor point on the joints and other key parts of the subject's body, and recover the parameters of the human body posture by capturing the position information of the cursor point during the movement of the human body.

Literature A. Kolahi, M. Hoviattalab, T. Rezaeian, M. Alizadeh, M. Bostan, H. Mokhtarzadeh. "Design of a marker-based human motion tracking system". Biomedical Signal Processing and Control, vol.2, pp.59 -67, February2007. A human body motion capture system based on cursor points is designed, and the method of statistical image RGB three-channel mean square error is used to process images, but the accuracy is not high and the universality is low.

Literature Figueroa P J, Leite N J, Barros R M L.A flexible software for tracking of markers used in human motion analysis[J].Computer Methods and Programs in Biomedicine,2003,72(2):155-165. Designed a cursor-based The human motion analysis software for points uses morphological processing methods to process images and extract cursor points. However, due to problems such as occlusion and irregular shapes of cursor points, the rate of misrecognition and missed recognition is high, and the obtained data is difficult to use for subsequent analysis and research.

Patent US8457382B2 proposes a system for identifying cursor points in X-ray images, including a scanner, a computer with an electronic processing unit and a memory unit for storing a cursor point recognition module. Image data is first generated with a scanner and transferred to a computer. The image is then high-pass filtered using an electronic processing unit to suppress the background, resulting in a background-suppressed image, and finally the cursor point is identified. This system is highly customized and expensive, making it difficult to use in common application scenarios.

In short, the existing cursor point capture and tracking technology mainly adopts methods based on image brightness or color. First, the cursor point area is segmented by using the brightness difference between the cursor point and background pixels or the color characteristics of the cursor point itself, and then the center of gravity of the area is Approximate as the center of the cursor point. This method is very sensitive to noise and strongly depends on image quality. Due to the occlusion problem during the shooting process, the cursor point will be lost in the middle frame. The pattern shape of the cursor point is often not regular enough, resulting in inaccurate positioning.

Contents of the invention

In view of this, it is necessary to provide a system and method for extracting and tracking a cursor point in an out-of-focus video.

The present invention provides a system for extracting and tracking a cursor point in an out-of-focus video, including an import module electrically connected to each other, a first frame image processing module, a segmentation module, an extraction module, a correction module and an update module, wherein: the The import module is used to import the first frame image of the out-of-focus video; the first frame image processing module is used to manually define one or more image areas as the search area of the cursor point in the first frame image, and record The endpoint coordinates of the search area, and each search area as a sub-image of the overall image respectively extract the cursor point and obtain the center coordinate of the cursor point; the import module is also used to import the next frame of the out-of-focus video Image; the segmentation module is used to segment sub-images in the next frame image according to the search area recorded in the previous frame image; the extraction module is used to extract the cursor point in each sub-image, and obtain the center coordinates of the cursor point ; The correction module is used to correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point; the update module is used to update the search area.

Wherein, the out-of-focus video is a human body movement video with a cursor point attached to it.

The extraction of the cursor point is to extract the cursor point by applying the Hough transform principle based on the gradient vector.

The segmentation module is specifically used to segment the next frame of image according to the endpoint coordinate values of the search area recorded in the previous frame of image, so as to obtain the search area in the next frame of image, that is, the sub-image.

The update module is specifically used to calculate the coordinate increment of the corrected cursor point center coordinate relative to the cursor point center coordinate in the previous frame image, and update the end point coordinate value of the search area according to the coordinate increment, thereby Update the search area.

The present invention also provides a method for extracting and tracking a cursor point in an out-of-focus video, the method comprising the following steps: a. Importing the first frame image of the out-of-focus video; b. Manually delimiting the first frame image One or more image areas are used as the search area of the cursor point, and the endpoint coordinate values of the search area are recorded, and each search area is used as a sub-image of the overall image to extract the cursor point and obtain the center coordinate of the cursor point; c. Import the next frame image of the out-of-focus video; d. segment sub-images in the next frame image according to the search area recorded in the previous frame image; e. extract the cursor point in each sub-image, and obtain the cursor point Center coordinates; f. Correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point g. Update the search area; h. Repeat steps c to g until the last frame of image.

Wherein, the out-of-focus video is a human body movement video with a cursor point attached to it.

The extraction of the cursor point is to extract the cursor point by applying the Hough transform principle based on the gradient vector.

The step d includes: segmenting the next frame of image according to the endpoint coordinate value of the search area recorded in the last frame of image, so as to obtain the search area in the next frame of image, that is, the sub-image.

The step g includes: calculating the coordinate increment of the corrected cursor point center coordinate relative to the cursor point center coordinate in the previous frame image, and updating the end point coordinate value of the search area according to the coordinate increment, thereby updating Search area.

The system and method for extracting and tracking the cursor point in the out-of-focus video provided by the present invention utilizes the characteristics of the cursor point in the out-of-focus video, and uses Hough transform to locate the center of the cursor point. The invention has low sensitivity to noise, can solve the problem of occlusion of the cursor point in the shooting process, effectively solves the problem of irregular shape of the cursor point pattern, and can accurately locate the center of the cursor point.

Description of drawings

Fig. 1 is the gradient direction schematic diagram of circular edge of the present invention;

Fig. 2 is the gradient vector schematic diagram of circular edge pixel point of the present invention;

Fig. 3 is a schematic diagram of the implementation environment of the present invention;

Fig. 4 is the functional block diagram of the system for extracting and tracking the cursor point in the out-of-focus video of the present invention;

FIG. 5 is a flow chart of the method for extracting and tracking a cursor point in an out-of-focus video according to the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings and specific embodiments.

In order to better explain the present invention, firstly, the basic principle of gradient vector-based Hough transform is described.

The Hough transform based on the gradient vector describes the graph in another mode, which traverses the parameter space, and maps the graph point set belonging to the mode to a point in the parameter space for detection. The traditional Hough transform has a large amount of calculation and requires high memory. The Hough transform based on the gradient vector has been widely used for its low computational load.

The gradient of an image f(x,y) at (x,y) coordinates is defined as follows:

$\begin{array}{ccc}<span\; class="notranslate">\; \&dtri;</span>\mathrm{<span\; class="notranslate">\; f</span>}& \mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; G</span>}& \frac{\mathrm{<span\; class="notranslate">\; f</span>}}{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; ,</span>\frac{\mathrm{<span\; class="notranslate">\; f</span>}}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}\end{array}$

The direction of the gradient vector of the image f(x,y) at the (x,y) coordinates is:

$\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& \mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}& \mathrm{<span\; class="notranslate">\; arctan</span>}& \frac{\mathrm{<span\; class="notranslate">\; G</span>}}{\mathrm{<span\; class="notranslate">\; G</span>}}\end{array}$

Referring to Figure 1, the direction of the gradient at the edge points of the circle is perpendicular to the tangent of the circle while passing through the center of the circle.

According to this feature, the gradient vectors of all pixels on the edge of the circle intersect at the center of the circle, as shown in Figure 2. Accumulate the pixels on the gradient vector of all pixels, and it is easy to find the peak point in the accumulator, which is the center coordinate of the cursor point.

Referring to FIG. 3 , it is a schematic diagram of the implementation environment of the present invention. The implementation environment of the present invention is mainly composed of a human body 10 , a camera 12 and a computer 16 .

Wherein, the human body 10 pastes a cursor point 14 on the part to be studied.

The camera 12 shoots out-of-focus video of human body movement, and transmits the out-of-focus video to the computer 16 .

The computer 16 may be an IBM-based computer (IBM Personal Computer, IBM PC), Apple's Mac PC, a personal computer, a network server, or any other suitable computer. A system for extracting and tracking the cursor point in the out-of-focus video is installed in the computer 16, and the system is used for extracting and tracking the cursor point in the out-of-focus video input by the above-mentioned camera 12.

Referring to FIG. 4 , it is a functional block diagram of the system for extracting and tracking the cursor point in the out-of-focus video installed in the computer 16 in FIG. 3 . The system includes an import module, a first frame image processing module, a segmentation module, an extraction module, a correction module and an update module.

The import module is used to import the first frame of image. Specifically, the import module imports the first frame image of the out-of-focus video of human body movement captured by the camera 12 .

The first frame image processing module is used to manually define one or more image areas as the search area of the cursor point in the first frame image, and record the endpoint coordinate values of the search area; Regions are used as sub-images of the overall image to extract the cursor point and obtain the center coordinates of the cursor point by applying the gradient vector-based Hough transform principle respectively. Wherein, there may be one cursor point or multiple cursor points in each search area.

The import module is also used to import the next frame image of the out-of-focus video.

The segmentation module is used to segment the sub-image in the next frame of image according to the search area recorded in the previous frame of image. Specifically, the segmenting module segments the image of the next frame according to the endpoint coordinate values of the search area recorded in the image of the previous frame, so as to obtain the sub-image which is the search area in the image of the next frame.

The extraction module is used to extract the cursor point in each sub-image by applying the Hough transform principle based on the gradient vector, and obtain the center coordinate of the cursor point.

The optimization module is used to apply an optimization algorithm, such as Kalman filter, particle filter, etc., to correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point.

The update module is used to calculate the coordinate increment of the corrected cursor point center coordinate relative to the cursor point center coordinate in the previous frame image, and update the end point coordinate value of the search area according to the coordinate increment, thereby updating the search area so that the sub-image always contains the cursor point to be tracked.

Referring to FIG. 5 , it is a flow chart of a preferred embodiment of the method for extracting and tracking a cursor point in an out-of-focus video according to the present invention.

Step S401, the import module imports the first frame of image. Specifically, the first frame image of the out-of-focus video of human body movement captured by the camera 12 is imported.

In step S402, the first frame image processing module manually defines one or more image areas in the first frame image as the search area for the cursor point, and records the coordinate values of the endpoints of the search area. The first frame image processing module uses each search area as a sub-image of the overall image to extract the cursor point by applying the Hough transform principle based on the gradient vector, and obtains the center coordinate of the cursor point. Wherein, there may be one cursor point or multiple cursor points in each search area.

Step S403, the import module imports the next frame image of the out-of-focus video.

Step S404, the extraction module divides sub-images in the next frame of image according to the search area recorded in the previous frame of image. Specifically, the next frame of image is segmented according to the endpoint coordinate values of the search area recorded in the previous frame of image, so as to obtain the search area in the next frame of image, that is, the sub-image.

In step S405, the optimization module extracts the cursor point from each sub-image by applying the Hough transform principle based on the gradient vector, and obtains the center coordinate of the cursor point.

Step S406, applying an optimization algorithm, such as Kalman filter, particle filter, etc., to correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point.

Step S407, the update module calculates the coordinate increment of the corrected cursor point center coordinate relative to the cursor point center coordinate in the previous frame image, and updates the end point coordinate value of the search area according to the coordinate increment, thereby updating The search area such that the sub-image always contains the cursor point to be tracked.

Step S403 to step S407 are repeated until the last frame of image.

Although the present invention has been described with reference to the current preferred embodiments, those skilled in the art should understand that the above-mentioned preferred embodiments are only used to illustrate the present invention, and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and scope of principles shall be included in the protection scope of the present invention.

Claims (10)

1. A system for extracting and tracking a cursor point in an out-of-focus video, characterized in that the system includes an electrically connected import module, a first frame image processing module, a segmentation module, an extraction module, a correction module and an update module ,in: The import module is used to import the first frame image of the out-of-focus video; The first frame image processing module is used to manually define one or more image areas as the search area of the cursor point in the first frame image, and record the endpoint coordinate values of the search area, and each search Regions are used as sub-images of the overall image to extract the cursor point and obtain the center coordinates of the cursor point; The import module is also used to import the next frame image of the out-of-focus video; The segmentation module is used to segment sub-images in the next frame image according to the search area recorded in the previous frame image; The extraction module is used to extract the cursor point in each sub-image, and obtain the center coordinates of the cursor point; The correction module is used to correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point; The updating module is used for updating the search area. 2. The system according to claim 1, characterized in that, the out-of-focus video is a human motion video taken with a cursor point attached. 3. The system according to claim 1, wherein said extracting the cursor point is to extract the cursor point by applying the Hough transform principle based on the gradient vector. 4. The system according to claim 1, wherein the segmentation module is specifically used to segment the next frame image according to the endpoint coordinate value of the search area recorded in the previous frame image, so as to obtain the next frame A search area in an image is a sub-image. 5. The system according to claim 1, wherein the update module is specifically used to calculate the coordinate increment of the corrected cursor point center coordinates relative to the cursor point center coordinates in the previous frame image, The coordinate value of the end point of the search area is updated according to the coordinate increment, thereby updating the search area. 6. A method for extracting and tracking a cursor point in an out-of-focus video, characterized in that the method comprises the steps: a. Import the first frame image of the out-of-focus video; b. Manually define one or more image areas in the first frame image as the search area of the cursor point, and record the endpoint coordinate values of the search area, and use each search area as a sub-image of the overall image Extract the cursor point separately and obtain the center coordinate of the cursor point; c. Import the next frame image of the out-of-focus video; d. Segment sub-images in the next frame image according to the search area recorded in the previous frame image; e. Extract the cursor point in each sub-image, and obtain the center coordinates of the cursor point; f. Correct the center coordinates of the cursor point obtained above, and output the corrected center coordinates of the cursor point g. Update the search area; h. Repeat step c to step g until the last frame of image. 7. The method according to claim 6, characterized in that, the out-of-focus video is a human body movement video taken with a cursor point attached. 8. The method according to claim 6, characterized in that said extracting the cursor point is to extract the cursor point by applying the Hough transform principle based on the gradient vector. 9. The method according to claim 6, wherein said step d comprises: segmenting the next frame of image according to the endpoint coordinate value of the search area recorded in the previous frame of image, so as to obtain the next frame of image The search area in is the sub-image. 10. The method according to claim 6, wherein said step g comprises: calculating the coordinate increment of the corrected cursor point center coordinates relative to the cursor point center coordinates in the previous frame image, according to The coordinate increment updates the coordinate value of the end point of the search area, thereby updating the search area.

Images