CN106920217A - The method and device of image flame detection - Google Patents

Abstract

A method and device for image correction, the method comprising: acquiring a first pixel value of a current phase point to be corrected in an original image; determining a correction multiple corresponding to the current phase point to be corrected; combining the first pixel value with The product of the correction multiple is used as the corrected pixel value of the current phase point to be corrected. The above scheme can improve the probability of obtaining a correct output image.

Description

Method and device for image correction

technical field

The invention relates to the technical field of image processing, in particular to an image correction method and device.

Background technique

With the continuous development of camera focus technology, focus technology has developed from manual focus to the current automatic focus. There is a focusing method in autofocus called Phase Detection AutoFocus (PDAF). In order to achieve phase focusing, the input light of some image pixels is partially blocked, and these pixels are called phase pixels or phase points (called phase points in this paper). is the phase point), the pixel that is blocked in the left half is called the left phase point, and the pixel that is blocked in the right half is called the right phase point.

In the phase focusing method, part of the incident light at the phase point is blocked, which will produce obvious images in the final image output, which is reflected in the phase point and its surrounding pixels being darker or brighter, or false colors appear. Therefore, it is necessary to correct the pixel value of the phase point to obtain a correct output image.

In the existing method of correcting the pixel value of a phase point, the phase point is usually corrected as a bad point, so that the pixel value of the phase point is almost completely lost, so that the reference pixel value data used for calculation during imaging is less, and errors occur. The probability is high, in other words, the probability of getting the correct output image is low.

Contents of the invention

The technical problem solved by the present invention is to provide an image correction method and device to correct the pixel value of the phase point, thereby increasing the probability of obtaining a correct output image.

In order to solve the above technical problems, an embodiment of the present invention provides an image correction method, the method comprising:

Obtain the first pixel value of the current phase point to be corrected in the original image;

Determine the correction multiple corresponding to the current phase point to be corrected;

The product of the first pixel value and the correction multiple is used as the corrected pixel value of the current phase point to be corrected.

Optionally, the correction multiple is determined in the following manner:

Let the camera module image the specified scene to obtain the calibration image;

The correction factor is calculated according to the pixel value of the pixel having the same color filter as the current phase point to be corrected in the calibration image and the first pixel value.

Optionally, the specified scene is a flat scene under uniform illumination.

Optionally, the calculating the correction multiple according to the pixel value of the pixel having the same color filter as the current phase point to be corrected in the calibration image and the first pixel value includes:

Acquiring the second pixel value of the current phase point to be corrected from the calibration image;

In the first neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixels having the same color filter as the current phase point to be corrected as the first average value;

The quotient of the first average value and the second pixel value is used as the correction factor.

Optionally, the image correction method also includes:

When the filter color of the current phase point to be corrected is different from the filter color of the pixels of the same color channel in the second neighborhood, refer to the channel color of the pixels of the same color channel in the second neighborhood and the The spectral correlation of the filter color of the phase point to be corrected is updated to calculate the corrected pixel value.

Optionally, updating and calculating the corrected pixel value with reference to the spectral correlation between the channel color of the pixel points of the same color channel in the second neighborhood and the filter color of the phase point to be corrected includes:

In the second neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixels that are the same as the color channel of the current phase point to be corrected as the second average value;

In the second neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixel points with the same color filter as the current phase point to be corrected, as a third average value;

using the difference between the third average value and the second average value as the spectral correlation;

The corrected pixel value is subtracted from the spectral correlation to update the corrected pixel value.

Optionally, the image correction method also includes:

Gaussian filtering is performed on the corrected pixel value of the phase point to be corrected and the pixel values of the pixel points with the same color channel in the second neighborhood, and the corrected pixel value is updated using the result of Gaussian filtering.

Optionally, the original image is a Bayer image.

An embodiment of the present invention also provides an image correction device, including:

The first pixel value acquisition unit is adapted to acquire the first pixel value of the current phase point to be corrected in the original image;

A correction multiple determination unit, adapted to determine the correction multiple corresponding to the current phase point to be corrected;

The corrected pixel value calculation unit is adapted to use the product of the first pixel value and the correction multiple as the corrected pixel value of the current phase point to be corrected.

Optionally, the correction multiple determination unit includes:

The calibration image acquisition subunit is adapted to acquire the calibration image obtained after the camera module images a specified scene;

The correction multiple calculation subunit is adapted to calculate the correction multiple according to the pixel value of the pixel in the calibration image having the same color filter as the current phase point to be corrected and the first pixel value.

Optionally, the specified scene is a flat scene under uniform illumination.

Optionally, the correction multiple calculation subunit includes:

A second pixel value acquisition module, adapted to acquire the second pixel value of the current phase point to be corrected from the calibration image;

The first average value calculation sub-module is adapted to calculate the average value of the pixel values of the pixels having the same color filter as the current phase point to be corrected in the first neighborhood of the current phase point to be corrected as the first average value;

The correction factor calculation sub-module is adapted to use the quotient of the first average value and the second pixel value as the correction factor.

Optionally, the image correction device also includes:

The correction pixel value correction unit is adapted to refer to the same color channel in the second neighborhood when the filter color of the current phase point to be corrected is different from the filter color of the pixel points of the same color channel in the second neighborhood The spectral correlation between the channel color of the pixel point and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value.

Optionally, the corrected pixel value correction unit includes:

The second average value calculation subunit is adapted to calculate the average value of the pixel values of the pixels in the same color channel as the current phase point to be corrected in the second neighborhood of the current phase point to be corrected as the second average value;

The third average calculation subunit is adapted to calculate the average value of the pixel values of the pixel points with the same color filter as the current phase point to be corrected in the second neighborhood of the current phase point to be corrected, as third average;

a spectral correlation calculation subunit, adapted to use the difference between the third average value and the second average value as the spectral correlation;

The corrected pixel value correction subunit is adapted to make a difference between the corrected pixel value and the spectral correlation, so as to update the corrected pixel value.

Optionally, the image correction device also includes:

The filtering unit is adapted to perform Gaussian filtering on the corrected pixel value of the phase to be corrected and the pixel values of the same color channel in the second domain, and update the corrected pixel value using the Gaussian filtering result.

Optionally, the original image is a Bayer image.

Compared with the prior art, the technical solutions of the embodiments of the present invention have the following beneficial effects:

The technical solution of the embodiment of the present invention obtains the first pixel value of the current phase point to be corrected in the original image, determines the correction multiple corresponding to the current phase point to be corrected, and calculates the product of the first pixel value and the correction multiple As the corrected pixel value of the current phase point to be corrected, the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the correction method that almost loses the phase point pixel value in the prior art. The problem of the wrong output image, thereby improving the probability of obtaining the correct output image.

Further, in the technical solution of the embodiment of the present invention, when the filter color of the current phase point to be corrected is different from the filter color of the pixel points of the same color channel in the second neighborhood, refer to the same color channel in the second neighborhood The spectral correlation between the channel color of the pixel point of the channel and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value, so that the calculation result of the corrected pixel value is more accurate.

Further, in the technical solution of the embodiment of the present invention, Gaussian filtering is performed on the corrected pixel value of the phase point to be corrected and the pixel value of the pixel point with the same color channel in the second neighborhood, so as to update the corrected pixel value , so that the pixel value of the corrected phase point is more accurate, thereby increasing the probability of obtaining a correct output image.

Description of drawings

Fig. 1 is a flow chart of a method for image correction in an embodiment of the present invention;

Fig. 2 is a flowchart of another image correction method in an embodiment of the present invention;

3A and 3B are Bayer images of the current phase point to be corrected and its neighborhood 5×7 window in the embodiment of the present invention;

Fig. 4 is a schematic structural diagram of an image correction device in an embodiment of the present invention;

Fig. 5 is a schematic structural diagram of a correction multiple determination unit in Fig. 4;

Fig. 6 is a schematic structural diagram of another image correction device in an embodiment of the present invention.

detailed description

As mentioned in the background technology, in the existing methods of correcting the pixel value of a phase point, the phase point is usually corrected as a bad point, so that the pixel value of the phase point is almost completely lost, so that the reference pixel used for calculation during imaging The value data is less, the chance of error is high, in other words, the probability of obtaining the correct output image is low.

The inventors of the present application have found through research that for a phase point, its occluder has been fixed, and the proportion of incident light blocked by it is constant, so it can be inferred that the phase point is multiplied by a certain fixed magnification to obtain the phase point imaging. The output pixel value. Therefore, the technical solution of the embodiment of the present invention obtains the first pixel value of the current phase point to be corrected in the original image, determines the correction multiple corresponding to the current phase point to be corrected, and compares the first pixel value with the correction multiple The product of is used as the corrected pixel value of the current phase point to be corrected, so that the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the correction that almost loses the phase point pixel value in the prior art The problem of error in the output image caused by the method, thereby improving the probability of obtaining a correct output image.

In order to make the above objects, features and beneficial effects of the present invention more comprehensible, specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

Fig. 1 is a flowchart of an image correction method in an embodiment of the present invention. The method for image correction shown in FIG. 1 will be described in detail below.

Step S101: Obtain the first pixel value of the current phase point to be corrected in the original image.

In a specific implementation, the original image is an image obtained by converting the light source signal captured by the image sensor into a digital signal. The original image records the original information of the image sensor. Usually, the original image inside the camera is a Bayer image. The general suffix is .raw, which contains the numerical information of the phase point. Of course, the original image can also be in other appropriate formats, as long as the phase point information is still preserved therein.

Step S102: Determine the correction multiple corresponding to the current phase point to be corrected.

In a specific implementation, the correction multiple is pre-calculated and stored, and each phase point to be corrected has a corresponding correction multiple, and the pre-stored correction multiple can be found through the corresponding relationship. Wherein, the correction multiple can be calculated by various methods.

Step S103: Use the product of the first pixel value and the correction factor as the corrected pixel value of the current phase point to be corrected.

The corrected pixel value is the corrected pixel value of the phase point to be corrected.

In an embodiment of the present invention, the correction multiple described in step S102 can be determined in the following manner:

Let the camera module image the specified scene to obtain the calibration image;

The correction factor is calculated according to the pixel value of the pixel having the same color filter as the current phase point to be corrected in the calibration image and the first pixel value.

In an example of the above method, the specified scene may be a flat scene under uniform illumination.

In an example of the above method, after the calibration image in the flat scene is obtained, according to the pixel value of the pixel in the calibration image that has the same color filter as the current phase point to be corrected, and the first The pixel value is used to calculate the correction factor, which may include:

Acquiring the second pixel value of the current phase point to be corrected from the calibration image;

In the first neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixels having the same color filter as the current phase point to be corrected as the first average value;

The quotient of the first average value and the second pixel value is used as the correction factor.

In a specific implementation, the size of the first neighborhood can be set as required, and there is no limitation here.

In this embodiment, by obtaining the first pixel value of the current phase point to be corrected in the original image, the correction multiple corresponding to the current phase point to be corrected is determined, and the product of the first pixel value and the correction multiple is used as the current The corrected pixel value of the phase point to be corrected, so that the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the error in the output image caused by the correction method that almost loses the phase point pixel value in the prior art problem, thereby increasing the probability of obtaining the correct output image.

Another embodiment of the present invention will be described below with reference to FIG. 2 .

Fig. 2 is a flowchart of another image correction method in an embodiment of the present invention. The image correction method will be described in detail below with reference to FIG. 2 .

Step S201: Obtain the first pixel value of the current phase point to be corrected in the original image;

Step S202: Determine the correction multiple corresponding to the current phase point to be corrected;

Step S203: taking the product of the first pixel value and the correction multiple as the corrected pixel value of the current phase point to be corrected;

Step S204: When the filter color of the current phase point to be corrected is different from the filter color of the pixels of the same color channel in the second neighborhood, refer to the channel of the pixels of the same color channel in the second neighborhood The spectral correlation between the color and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value;

Step S205: Perform Gaussian filtering on the corrected pixel value of the phase point to be corrected and the pixel values of the pixel points with the same color channel in the second neighborhood, and update the corrected pixel value using the Gaussian filtering result.

In this embodiment, for the description of step S201 to step S203, please refer to the description of step S101 to step S103 in FIG. 1 correspondingly, and details are not repeated here.

In the specific implementation of step S204, the corrected pixel value may be updated and calculated in the following manner:

In the second neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixels that are the same as the color channel of the current phase point to be corrected as the second average value;

In the second neighborhood of the current phase point to be corrected, calculate the average value of the pixel values of the pixel points with the same color filter as the current phase point to be corrected, as a third average value;

using the difference between the third average value and the second average value as the spectral correlation;

The corrected pixel value is subtracted from the spectral correlation to update the corrected pixel value.

The implementation manner of the above step S204 is illustrated below with reference to FIG. 3A and FIG. 3B :

Figures 3A and 3B show a bayer image of a 5x7 window of a phase point and its neighborhood. Figure 3A specifically shows the color channel of each pixel in the bayer image, R indicates the red channel (R channel), G indicates the green channel (G channel), and B indicates the blue channel (B channel), which is currently to be corrected Please refer to the shaded part in Figure 3A for the phase point o, and Figure 3B specifically identifies the current phase point o to be corrected and the points a, b, c, d, e, f, g, and h in its 3×3 neighborhood .

Please compare Figure 3A and Figure 3B, the current phase point o to be corrected is in the G channel, and there are also points a, b, c and d in the G channel in its 3×3 neighborhood, and points e and f in the figure are at R channel, g point and h point are in B channel.

The second neighborhood in this embodiment is the 3×3 neighborhood of the current phase point o to be corrected. Usually, the filter color on the pixel point cover is the same as the channel color of the pixel point, but the phase The filter color overlaid on the point may be different from the channel color described by the phase point. As shown in Figure 3A and Figure 3B, the filter colors of points a, b, c, and d are the same as the channel colors, all of which are red, while the filter colors of the phase points may be other colors at this time, assuming blue in this example color. At this time, the corrected pixel value may be updated and calculated with reference to the spectral correlation between the channel color of the pixel points of the same color channel in the second neighborhood and the filter color of the phase point to be corrected.

Specifically, within the 3×3 neighborhood of the current phase point o to be corrected, calculate the pixel values of the pixel points (points a, b, c, and d) that are the same as the color channel of the current phase point o to be corrected Average value, as the second average value m;

In the 3×3 neighborhood of the current phase point o to be corrected, calculate the average value of the pixel values of the pixel points (points e and f) identical to the filter color blue of the current phase point o to be corrected, to as the third average n;

Using the difference n-m between the third average value and the second average value as the spectral correlation r;

Assuming that the corrected pixel value calculated in step S203 is T, the corrected pixel value T is subtracted from the spectral correlation r, and the corrected pixel value T is updated using T-r.

Please continue to refer to FIG. 3B. In the specific implementation of step S205, the corrected pixel value of the phase point o to be corrected is compared with the pixel points (points a, b, c, and d) having the same color channel in the second neighborhood. Gaussian filtering is performed on the pixel values, and the corrected pixel values are updated using the Gaussian filtering results.

It should be pointed out that referring to FIG. 1 , after obtaining the corrected pixel value in step S103 , the method described in step S204 in this embodiment may also be used for filtering to update the corrected pixel value obtained in step S103 .

In this embodiment, by obtaining the first pixel value of the current phase point to be corrected in the original image, the correction multiple corresponding to the current phase point to be corrected is determined, and the product of the first pixel value and the correction multiple is used as the current The corrected pixel value of the phase point to be corrected, so that the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the error in the output image caused by the correction method that almost loses the phase point pixel value in the prior art problem, thereby increasing the probability of obtaining the correct output image.

Further, in this embodiment, when the filter color of the current phase point to be corrected is different from the filter color of the pixel points of the same color channel in the second neighborhood, refer to the pixel points of the same color channel in the second neighborhood The spectral correlation between the channel color of the channel color and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value, so that the calculation result of the corrected pixel value is more accurate.

Further, in this embodiment, Gaussian filtering is performed on the corrected pixel value of the phase point to be corrected and the pixel value of the pixel point with the same color channel in the second neighborhood to update the corrected pixel value, so that the corrected The pixel values of the final phase points are more accurate, thereby increasing the probability of obtaining a correct output image.

The device corresponding to the image correction method in the embodiment of the present invention will be further described in detail below with reference to FIG. 4 and FIG. 5 .

Fig. 4 is a schematic structural diagram of an image correction device in an embodiment of the present invention, please refer to Fig. 4, the image correction device 40 may include:

The first pixel value acquiring unit 401 is adapted to acquire the first pixel value of the current phase point to be corrected in the original image;

The correction multiple determination unit 402 is adapted to determine the correction multiple corresponding to the current phase point to be corrected;

The corrected pixel value calculation unit 403 is adapted to use the product of the first pixel value and the correction multiple as the corrected pixel value of the current phase point to be corrected.

In a specific implementation, the correction multiple determination unit 402 may include a calibration image imaging subunit 4021 and a correction multiple calculation subunit 4022, please refer to FIG. 5 , which shows the correction multiple determination unit described in FIG. 4 A structural schematic diagram of FIG. 5 , the correction multiple determination unit 402 having the structure in FIG. 5 can pre-calculate the correction multiple. Or, in another specific implementation, the correction multiple determination unit 402 may also only have a search function, and acquire previously determined correction multiples from an external or internal database, memory, or the like.

The calibration image acquisition subunit 4021 is adapted to acquire a calibration image obtained by imaging a specified scene in the camera module;

The correction multiple calculation subunit 4022 is adapted to calculate the correction multiple according to the pixel value of the pixel in the calibration image having the same color filter as the current phase point to be corrected and the first pixel value.

In a specific implementation, the specified scene may be a flat scene under uniform illumination.

In a specific implementation, the correction multiple calculation subunit 4022 may include:

A second pixel value acquisition module, adapted to acquire the second pixel value of the current phase point to be corrected from the calibration image;

The first average value calculation sub-module is adapted to calculate the average value of the pixel values of the pixels having the same color filter as the current phase point to be corrected in the first neighborhood of the current phase point to be corrected as the first average value;

The correction factor calculation sub-module is adapted to use the quotient of the first average value and the second pixel value as the correction factor.

In this embodiment, by obtaining the first pixel value of the current phase point to be corrected in the original image, the correction multiple corresponding to the current phase point to be corrected is determined, and the product of the first pixel value and the correction multiple is used as the current The corrected pixel value of the phase point to be corrected, so that the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the error in the output image caused by the correction method that almost loses the phase point pixel value in the prior art problem, thereby increasing the probability of obtaining the correct output image.

Figure 6 is a schematic structural diagram of an image correction device in an embodiment of the present invention, please refer to Figure 6, the image correction device 60 may include: a first pixel value acquisition unit 601, a correction factor determination unit 602 and a correction pixel Value calculation unit 603.

Wherein, for the description of the first pixel value acquisition unit 601, the correction multiple determination unit 602 and the corrected pixel value calculation unit 603, please refer to the first pixel value acquisition unit 401, the correction multiple determination unit 402 and the corrected pixel value in FIG. The description of the value calculation unit 403 is omitted here.

The image correcting apparatus 60 of this embodiment may further include: a corrected pixel value modifying unit 604 and a filtering unit 605 . in:

The corrected pixel value correction unit 604 is adapted to refer to the original image when the filter color of the current phase point to be corrected is different from the filter color of the pixels of the same color channel in the second neighborhood. The spectral correlation between the channel color of the pixel points of the same color channel in the second neighborhood and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value.

The filtering unit 605 is adapted to perform Gaussian filtering on the corrected pixel value of the phase to be corrected and the pixel value of the pixel point of the same color channel in the second domain, and update the corrected pixel value using the result of Gaussian filtering .

In a specific implementation, the corrected pixel value correction unit 604 may include:

The second average value calculation subunit 6041 is adapted to calculate the average value of the pixel values of the pixels in the same color channel as the current phase point to be corrected in the second neighborhood of the current phase point to be corrected, so as to as the second average;

The third average value calculation subunit 6042 is adapted to calculate the average value of the pixel values of the pixel points with the same color filter as the current phase point to be corrected in the second neighborhood of the current phase point to be corrected, so as to as the third average;

The spectral correlation calculation subunit 6043 is adapted to use the difference between the third average value and the second average value as the spectral correlation;

The corrected pixel value modifying subunit 6044 is adapted to make a difference between the corrected pixel value and the spectral correlation, so as to update the corrected pixel value.

In this embodiment, by obtaining the first pixel value of the current phase point to be corrected in the original image, the correction multiple corresponding to the current phase point to be corrected is determined, and the product of the first pixel value and the correction multiple is used as the current The corrected pixel value of the phase point to be corrected, so that the corrected phase point pixel value can be used to participate in the calculation in the subsequent processing of the image, thereby avoiding the error in the output image caused by the correction method that almost loses the phase point pixel value in the prior art problem, thereby increasing the probability of obtaining the correct output image.

Further, in this embodiment, when the filter color of the current phase point to be corrected is different from the filter color of the pixel points of the same color channel in the second neighborhood, refer to the pixel points of the same color channel in the second neighborhood The spectral correlation between the channel color of the channel color and the filter color of the phase point to be corrected is updated to calculate the corrected pixel value, so that the calculation result of the corrected pixel value is more accurate.

Further, in this embodiment, Gaussian filtering is performed on the corrected pixel value of the phase point to be corrected and the pixel value of the pixel point with the same color channel in the second neighborhood to update the corrected pixel value, so that the corrected The pixel values of the final phase points are more accurate, thereby increasing the probability of obtaining a correct output image.

Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above-mentioned embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the storage medium can include: ROM , RAM, disk or CD, etc.

The method and system of the embodiment of the present invention have been introduced in detail above, and the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, so the protection scope of the present invention should be based on the scope defined in the claims.

Claims (16)

1. a kind of method of image flame detection, it is characterised in that including：

Obtain the first pixel value of current phase point to be corrected in original image；

Determine the corresponding correction multiple of the phase point currently to be corrected；

Using first pixel value and the product for correcting multiple as the strong of the current phase point to be corrected Positive pixel value.

2. the method for image flame detection according to claim 1, it is characterised in that the correction multiple passes through In the following manner determines：

Allow camera module to be imaged given scenario, obtain uncalibrated image；

According to the picture of the pixel with the current phase point to be corrected in the uncalibrated image with homochromy filter Element value, and correct multiple described in first calculated for pixel values.

3. the method for image flame detection according to claim 2, it is characterised in that the given scenario is equal Planar scene under even illumination.

4. the method for image flame detection according to claim 3, it is characterised in that described according to the demarcation There is the pixel value of the pixel of homochromy filter, Yi Jisuo in image with the current phase point to be corrected State and multiple is corrected described in the first calculated for pixel values, including：

The second pixel value of the current phase point to be corrected is obtained from the uncalibrated image；

In the first neighborhood of the current phase point to be corrected, calculate and have with the current phase point to be corrected There is the average value of pixel value of the pixel of homochromy filter as the first average value；

Using the business of first average value and second pixel value as the correction multiple.

5. the method for image flame detection according to claim 1, it is characterised in that also include：

When the pixel in the filter color and the second neighborhood of the current phase point to be corrected with Color Channel When filter color is different, with reference to the passage face of the pixel of the same Color Channel in second neighborhood Color updates with the spectral correlations of the filter color of the phase point to be corrected and calculates correction pixel value.

6. the method for image flame detection according to claim 5, it is characterised in that the neighborhood of the reference second The passage color of the pixel of the interior same Color Channel and the filter color of the phase point to be corrected Spectral correlations update and calculate correction pixel value, including：

In the second neighborhood of the current phase point to be corrected, calculate and the current phase point to be corrected The average value of Color Channel identical pixel pixel value, as the second average value；

In the second neighborhood of the current phase point to be corrected, calculate and the current phase point to be corrected The average value of the pixel value of the color filter identical pixel, as the 3rd average value；

Using the difference of the 3rd average value and second average value as the spectral correlations；

The correction pixel value is made the difference with the spectral correlations, to update the correction pixel value.

7. the method for the image flame detection according to claim 5 or 6, it is characterised in that also include：

To there is same color passage in correction pixel value and second neighborhood of the phase point to be corrected The pixel value of pixel carries out gaussian filtering, and uses the result renewal correction pixel of gaussian filtering Value.

8. the method for the image flame detection according to claim any one of 1-6, it is characterised in that described original Image is Bayer images.

9. a kind of device of image flame detection, it is characterised in that including：

First pixel value acquiring unit, is suitable to obtain the first pixel of current phase point to be corrected in original image Value；

Correction multiple determining unit, is adapted to determine that the corresponding correction multiple of the phase point currently to be corrected；

Correction pixel value calculating unit, be suitable to using first pixel value with it is described correction multiple product as The correction pixel value of the current phase point to be corrected.

10. the device of image flame detection according to claim 9, it is characterised in that the correction multiple determines Unit, including：

Uncalibrated image obtains subelement, is suitable to obtain the demarcation obtained after camera module is imaged given scenario Image；

Correction multiple computation subunit, be suitable to according in the uncalibrated image with the current phase point to be corrected The pixel value of the pixel with homochromy filter, and correct multiple described in first calculated for pixel values.

The device of 11. image flame detections according to claim 10, it is characterised in that the given scenario is equal Planar scene under even illumination.

The device of 12. image flame detections according to claim 11, it is characterised in that the correction multiple is calculated Subelement, including：

Second pixel value acquisition module, is suitable to obtain the current phase point to be corrected from the uncalibrated image The second pixel value；

First mean value calculation submodule, is suitable in the first neighborhood of the current phase point to be corrected, meter The average value of pixel value of the pixel that there is homochromy filter with the current phase point to be corrected is calculated to make It is the first average value；

Correction multiple calculating sub module, be suitable to using the business of first average value and second pixel value as The correction multiple.

The device of 13. image flame detections according to claim 9, it is characterised in that also include：

Correction pixel value modification unit, is suitable to the filter color when the current phase point to be corrected and the second neighbour When in domain with the filter color difference of the pixel of Color Channel, with reference to described same in second neighborhood The spectral correlation of the passage color of the pixel of Color Channel and the filter color of the phase point to be corrected Property update calculate correction pixel value.

The device of 14. image flame detections according to claim 13, it is characterised in that the correction pixel value is repaiied Positive unit includes：

Second mean value calculation subelement, is suitable in the second neighborhood of the current phase point to be corrected, meter The average value with the Color Channel identical pixel pixel value of the current phase point to be corrected is calculated, with As the second average value；

3rd mean value calculation subelement, is suitable in the second neighborhood of the current phase point to be corrected, meter The average value with the pixel value of the color filter identical pixel of the current phase point to be corrected is calculated, As the 3rd average value；

Spectral correlations computation subunit, is suitable to make the difference of the 3rd average value and second average value It is the spectral correlations；

Correction pixel value modification subelement, is suitable to make the difference the correction pixel value with the spectral correlations, To update the correction pixel value.

The device of 15. image flame detection according to claim 13 or 14, it is characterised in that also include：

Filter unit, is suitable to the correction pixel value of the phase to be corrected and same color in second field The pixel value of the pixel of passage carries out gaussian filtering, and described strong using the result renewal of gaussian filtering Positive pixel value.

The device of 16. image flame detection according to any one of claim 9-14, it is characterised in that the original Beginning image is Bayer images.