Mixed type purple edge correction method and system, electronic equipment and medium

Abstract

The invention provides a mixed type purple boundary correction method and system, electronic equipment and a medium. The method comprises the following steps: calculating corresponding purple boundary intensity based on basic information of pixel points in an original image and basic information of corresponding first candidate points; according to the purple edge intensity of each pixel point, determining whether the pixel point is a purple edge pixel point, and if not, skipping the current pixel point; if yes, acquiring an adaptive filtering window of the current purple edge pixel point by using a set image calibration strategy; calculating corresponding candidate compensation colors according to the self-adaptive filtering window and purple boundary intensity of each purple boundary pixel point; and according to the candidate compensation color of each purple edge pixel point, the basic information and the basic information of the corresponding second candidate pixel point, calculating to obtain a corresponding saturation reduction intensity, and according to the saturation reduction intensity of each purple edge pixel point, carrying out saturation reduction processing on the candidate compensation color to obtain a corresponding target compensation color. According to the method and the device, the purple edge can be effectively removed, and meanwhile, the processed image space is naturally transited.

Description

Technical Field

[0001] The present application belongs to the field of image processing technology, and in particular relates to a hybrid type purple fringing correction method, system, electronic device and medium. Background Art

[0002] During the shooting or imaging display process, due to lens chromatic aberration, imaging device crosstalk, and optical chromatic aberration, a purple-colored area will appear at the junction of the overexposed area and the normal exposed area. This is called purple fringing (PF). These purple-colored colors are not the color of the object itself, and purple fringing obviously affects the visual effect of the image. The appearance of purple fringing is related to the shooting scene, camera lens, sensor, and internal interpolation algorithm. Affected by the shooting scene, camera lens, sensor, and internal interpolation algorithm, the original image obtained by the shooting cannot avoid the appearance of purple fringing, which greatly affects the image quality and visual effect.

[0003] Changing the lens design can reduce the appearance of image purple fringing to a certain extent, but it is easy to have incomplete purple fringing removal and unnatural spatial transition in local areas of the image after purple fringing removal, and the design cost is high. In addition, due to the complexity of the shooting scene and the quality of the lens, various types of purple fringing are prone to appear, such as purple stripes that often appear at the junction of bright and dark areas, and purple fringing that is tens of pixels wide in non-highlight areas. Therefore, how to eliminate purple fringing is a problem that needs to be solved. Summary of the invention

[0004] The present application provides a hybrid type purple fringing correction method, system, electronic device and medium, which are used to solve the problems of incomplete purple fringing removal in the purple fringing correction method in the prior art and unnatural spatial transition in local areas of the image after purple fringing removal.

[0005] In a first aspect, the present application provides a hybrid type purple fringing correction method, the method comprising: obtaining a color original image; calculating the corresponding purple fringing intensity based on basic information of each pixel in the original image and basic information of the corresponding first candidate point; the basic information comprises brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; determining whether the current pixel point is a purple fringed pixel point according to the purple fringing intensity of each pixel point, if not, skipping the current pixel point; if it is a purple fringed pixel point, obtaining an adaptive filtering window of the current purple fringed pixel point by using a set image calibration strategy ... determining whether the current pixel point is a purple fringed pixel point according to the purple fringing intensity of each pixel point, if not, skipping the current pixel point; determining whether the current pixel point is a purple fringed pixel point according to the purple fringing intensity of each pixel point, if not, skipping the current pixel point; determining whether the current pixel point is a purple fringed pixel point, if not, skipping the current pixel point; if it is a purple f The adaptive filtering window of the edge pixel points and the corresponding purple-fringing intensity are used to calculate the candidate compensation color corresponding to each purple-fringed pixel point; the corresponding desaturation intensity is calculated according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point; the second candidate pixel point is other pixel points within the adaptive filtering window of the current purple-fringed pixel point; the corresponding original pixel points are corrected according to the target compensation color of each purple-fringed pixel point to obtain a color image after removing the purple fringing.

[0006] In the process of purple fringing detection, the present application fully considers the formation causes and distribution characteristics of mixed-type purple fringing, and uses the mixed dimension of space and brightness to calculate the purple fringing intensity of each pixel, so that the purple fringing detection intensity has a smooth spatial transition, so that the image after subsequent purple fringing correction can be smoothly connected with the normal area, avoiding the local mutation and unnatural transition of the processed image; in the process of purple fringing correction, the formation causes and distribution laws of mixed-type purple fringing are fully considered, and an adaptive filtering window is adaptively generated for each purple-fringed pixel point, and purple fringes of different types and distributions are removed through two steps of filtering and desaturation, further reducing the residual purple fringing, so that various types of purple fringing are removed more thoroughly, avoiding the situation of incomplete purple fringing removal. The present application can effectively remove purple fringing while ensuring that the spatial transition of the processed image is natural, and avoiding the situation of spatial discontinuity in chromaticity of the image after purple fringing removal.

[0007] In an implementation method of the first aspect, determining whether the current pixel is a purple-fringed pixel based on the purple-fringing intensity of each pixel includes: judging based on the purple-fringing intensity of each pixel, if the purple-fringing intensity of the current pixel is greater than 0, then the current pixel is a purple-fringed pixel; if the purple-fringing intensity of the current pixel is equal to 0, then the current pixel is not a purple-fringed pixel.

[0008] In an implementation of the first aspect, obtaining an adaptive filtering window for a current purple-fringed pixel point using a set image calibration strategy includes: calibrating the original image using the set image calibration strategy to obtain a set of purple-fringed offsets corresponding to purple-fringed pixel points at different positions in the original image; and adaptively generating an adaptive filtering window for the current purple-fringed pixel point based on the purple offset set corresponding to the current purple-fringed pixel point.

[0009] In an implementation of the first aspect, the candidate compensation color corresponding to each purple-fringed pixel is calculated based on the adaptive filtering window of each purple-fringed pixel and the corresponding purple-fringing intensity, including: obtaining the corresponding second candidate pixel according to the adaptive filtering window of each purple-fringed pixel; the second candidate pixel is other pixel within the adaptive filtering window of the current purple-fringed pixel; the corresponding brightness difference is calculated based on the brightness value of each purple-fringed pixel and the brightness value of the corresponding second candidate pixel; the chroma saturation score of each second candidate pixel corresponding to each purple-fringed pixel is calculated based on the chroma of each second candidate pixel corresponding to each purple-fringed pixel; the chroma of each purple-fringed pixel after filtering is calculated based on the brightness difference and chroma saturation score of each second candidate pixel corresponding to each purple-fringed pixel, as the candidate compensation color.

[0010] In an implementation of the first aspect, a corresponding desaturation intensity is calculated based on the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding first candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple-fringed pixel point to obtain a target compensation color corresponding to each purple-fringed pixel point, including: calculating a corresponding highlight high-contrast score based on the brightness value of each purple-fringed pixel point and the brightness value of the corresponding first candidate pixel point; the highlight high-contrast score is calculated by a highlight dimension score and a high-contrast dimension score; calculating a corresponding desaturation intensity based on the highlight high-contrast score of each purple-fringed pixel point, the candidate compensation color and a set high desaturation intensity; desaturating the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel point to obtain a target compensation color corresponding to each purple-fringed pixel point.

[0011] In an implementation method of the first aspect, in the purple fringing detection process, the corresponding purple fringing intensity is calculated based on the basic information of the pixel points in the original image and the basic information of the corresponding first candidate points, including: calculating the corresponding highlight dimension score based on the brightness value of each pixel point in the original image and the brightness value of the corresponding first candidate point; calculating the corresponding high-contrast dimension score based on the brightness value of each pixel point in the original image and the brightness value of the corresponding first candidate point; calculating the corresponding chroma saturation score based on the hue of each pixel point in the original image; and calculating the corresponding purple fringing intensity according to the highlight dimension score, the high-contrast dimension score and the chroma saturation score corresponding to each pixel point.

[0012] In an implementation of the first aspect, the corresponding highlight dimension score is calculated based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point, including: calculating the brightness difference score of each pixel based on the brightness value of the first candidate point corresponding to each pixel in the original image and a preset highlight threshold; calculating the highlight spatial intensity score of each pixel based on the distance between each pixel and each highlight pixel in the detection area; calculating the highlight pixel score of each pixel based on the brightness value of each pixel, the brightness value of the corresponding first candidate point and a preset highlight fluctuation range; and calculating the corresponding highlight dimension score according to the brightness difference score of each pixel, the highlight spatial intensity score and the highlight pixel score.

[0013] In an implementation of the first aspect, calculating the corresponding high contrast dimension score based on the brightness value of the pixel point in the original image and the brightness value of the corresponding first candidate point includes: calculating the brightness difference of each pixel point based on the brightness value of each pixel point in the original image and the brightness value of the corresponding first candidate pixel point; calculating the contrast difference score of each pixel point based on the brightness difference of each pixel point and a preset contrast threshold; calculating the contrast spatial intensity score of each pixel point based on the distance between each pixel point and each high contrast pixel point in the detection area; and calculating the corresponding high contrast dimension score according to the contrast difference score and the contrast spatial intensity score of each pixel point.

[0014] In an implementation of the first aspect, calculating the corresponding hue-saturation score based on the hue of the pixel in the original image includes: calculating the corresponding purple score based on the hue of the pixel in the original image, a set purple range, and a set purple transition range; calculating the corresponding saturation score based on the saturation of the pixel in the original image and a set saturation range; calculating the corresponding hue-saturation score based on the purple score and the saturation score of each pixel.

[0015] In a second aspect, the present application provides a hybrid type purple fringing correction system, the system comprising: an acquisition module, configured to acquire a color original image; a purple fringing intensity calculation module, configured to calculate the corresponding purple fringing intensity based on basic information of each pixel in the original image and basic information of the corresponding first candidate point; the basic information comprises brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; an adaptive filtering window construction module, configured to determine whether the current pixel point is a purple fringed pixel point according to the purple fringing intensity of each pixel point, if not, skip the current pixel point; if it is a purple fringed pixel point, use the set image calibration strategy to obtain the adaptive filtering window of the current purple fringed pixel point; the candidate correction module comprises a plurality of pixel points, a plurality of pixel points, a plurality of pixel points, and a plurality of pixel points. The compensation color calculation module is configured to calculate the candidate compensation color corresponding to each purple-fringed pixel point according to the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringed intensity; the target compensation color calculation module is configured to calculate the corresponding desaturation intensity according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and desaturate the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point; the second candidate pixel point is other pixel points within the adaptive filtering window of the current purple-fringed pixel point; the correction module is configured to correct the corresponding original pixel points according to the target compensation color of each purple-fringed pixel point to obtain a color image after removing the purple fringes.

[0016] In a third aspect, the present application provides an electronic device, comprising: a memory and a processor; the memory is configured to store a processor executable program; the processor is configured to execute the executable program stored in the memory, so that the electronic device performs the above-mentioned mixed type purple fringing correction method.

[0017] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by an electronic device, implements the above-mentioned hybrid type purple fringing correction method.

[0018] As described above, the hybrid type purple fringing correction method, system, electronic device and medium described in the present application have the following beneficial effects:

[0019] The present application calculates the corresponding purple-fringing intensity based on the basic information of each pixel in the original image and the basic information of the corresponding first candidate point; the basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; the adaptive filtering window of each purple-fringed pixel point is obtained by setting an image calibration strategy according to the purple-fringing intensity of each pixel point, and the candidate compensation color corresponding to each purple-fringed pixel point is calculated according to the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringing intensity; the corresponding desaturation intensity is calculated according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturated intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point; the second candidate pixel point is other pixel points in the adaptive filtering window of the current purple-fringed pixel point; the corresponding original pixel points are corrected according to the target compensation color of each purple-fringed pixel point to obtain a color image after removing the purple fringing. By fully considering the spatial transition of the detection and correction processes in the purple fringing detection and correction stages, it is possible to effectively remove the purple fringing while ensuring a natural spatial transition of the processed image and avoiding spatial discontinuity in chromaticity of the image after purple fringing removal.

[0020] In the judgment conditions of highlight and high contrast, the present application considers both the value range and the spatial domain. In the value range dimension, the brightness value of the highlight pixel itself, or the difference between the bright pixel and the current pixel, is considered; in the spatial domain dimension, the spatial distance from the highlight pixel or the bright pixel to the current pixel is considered. The judgment of highlight and high contrast is integrated in two dimensions to achieve a smooth spatial transition of the detection intensity. In the judgment conditions of color, the "strict purple area" and the "transitional purple area" are considered. The closer to the "strict purple area", the higher the color discrimination score; the farther away from the "strict purple area", the lower the color discrimination score; the point beyond the "transitional purple area", the color discrimination score is 0. It can further ensure the smooth spatial transition of the detected purple edge intensity.

[0021] In the process of compensating color correction in the present application, purple-fringed pixels of different intensities correspond to corrections of different intensities. The detected and corrected pixels and the normal pixels that are not detected have better spatial transition after processing, reducing the local jumps and unnatural images caused by previous purple-fringing removal methods. At the same time, in the process of desaturation processing, residual purple-fringed pixels of different intensities correspond to desaturation processing of different intensities. The desaturated pixels and the pixels that have not entered the desaturation processing are fused by gradient weights after processing, ensuring better spatial connection and reducing the local jumps and unnatural images caused by previous desaturation purple-fringing removal methods.

[0022] The present application performs purple fringing correction processing through two stages: compensating color processing and desaturation processing. In the compensating color processing stage, the original color of the pixel contaminated by purple fringing can be restored. In the desaturation processing stage, for the strong purple fringing and ultra-wide purple fringing that cannot be completely covered in the compensating color processing stage, the desaturation method is further adopted to reduce the residual purple fringing. In the compensating color processing stage, the joint value range filtering method is used, and the weight of the color component and the weight of the brightness difference are considered at the same time. The weight of the color component can filter out the purple as much as possible after filtering, and it can also be naturally blended and transitioned with the surrounding pixels; the brightness difference weight can effectively reduce the problems of cross-color and color overflow that may occur during the purple fringing correction process. In the desaturation processing stage, the degree of highlight and the spatial attenuation transition of the highlight radiation area are fully considered, and the attenuation transition is used as a guide to further desaturate the residual purple fringing around the highlight, ensuring the natural spatial transition between the desaturated area and the non-desaturated area. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Shown is a schematic diagram of a hardware application scenario of the hybrid type purple fringing correction described in an embodiment of the present application.

[0024] Figure 2 Shown is a flow chart of a hybrid type purple fringing correction method according to an embodiment of the present application.

[0025] Figure 3 Shown is a schematic diagram of the process of purple-edge pixel confirmation described in an embodiment of the present application.

[0026] Figure 4 Shown is a schematic diagram of the overall process of the hybrid type purple fringing correction method described in an embodiment of the present application.

[0027] Figure 5 Shown is a schematic diagram of the compensation color acquisition described in an embodiment of the present application.

[0028] Figure 6 Shown is a structural schematic diagram of a hybrid type purple fringing correction system described in an embodiment of the present application.

[0029] Figure 7 Shown is a schematic diagram of the structure of an electronic device described in an embodiment of the present application.

[0030] Component number description

[0031] 1 Chip

[0032] 11 Memory

[0033] 12 Processor

[0034] 3 Hybrid type purple fringing correction system

[0035] 31 Get Module

[0036] 32 Purple fringing intensity calculation module

[0037] 33 Adaptive Filter Window Building Block

[0038] 34 Candidate compensation color calculation module

[0039] 35 Target compensation color calculation module

[0040] 36 Calibration module

[0041] 4 Electronic devices

[0042] 41 Processing Units

[0043] 42 Memory

[0044] 421 RAM

[0045] 422 Cache memory

[0046] 423 Storage System

[0047] 424 Utilities

[0048] 4241 Program Module

[0049] 43 Bus

[0050] 44 I / O interfaces

[0051] 45 Network Adapter

[0052] Steps S1 to S6

[0053] Steps S1'~S5'

[0054] Steps S31 to S32 DETAILED DESCRIPTION

[0055] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0056] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application, and thus the drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed at will, and the component layout may also be more complicated.

[0057] The following embodiments of the present application provide a hybrid type purple fringing correction method, system, electronic device and medium, which solve the problems of incomplete purple fringing removal in the purple fringing correction method in the prior art and unnatural spatial transition in local areas of the image after purple fringing removal.

[0058] PF: PurpleFringe, abbreviation for purple fringing.

[0059] CAC: Color Aberration Correction, the abbreviation of chromatic aberration correction.

[0060] DPF: De-PurpleFringe, abbreviation of purple fringing removal module or function.

[0061] YUV domain: a color space used for image and video processing, where Y represents luminance, and U and V represent chrominance. The YUV domain separates luminance from color information, allowing luminance and chrominance information to be processed independently during video compression and image processing.

[0062] like Figure 1 As shown, this embodiment provides a hardware application scenario of hybrid type purple fringing correction, specifically including a chip 1, the chip 1 receives an original image input by an external device, performs purple fringing correction and outputs a purple fringed image. The chip 1 may include a memory 11 and a processor 12. The chip 1 may be connected to an external memory and / or communicate with an external device ( Figure 1 not shown). Figure 1 The chip 1 shown may include components associated with the current example. Therefore, it will be clear to those skilled in the art that the chip 1 may also include components other than Figure 1 Other common components besides those shown in .

[0063] Here, the chip 1 may be implemented using various types of devices such as a personal computer (PC), a server device, a mobile device, an embedded device, etc. In detail, the chip 1 may be included in a smart phone that can capture an image and / or process an image, a tablet device, an augmented reality (AR) device, an Internet of Things (IoT) device, an autonomous driving vehicle, a robotic device, or a medical device, but is not limited thereto.

[0064] The memory 11 stores various data processed in the chip 1. For example, the memory 11 may store data that has been processed or is to be processed in the chip 1. In one example, the memory 11 may store instructions executable in the processor 12. In addition, the memory 11 may store an application or driver to be driven by the chip 1.

[0065] For example, the memory 11 may include a random access memory (RAM) such as a dynamic random access memory (DRAM) or a static random access memory (SRAM), a read-only memory (RAM), an electrically erasable programmable read-only memory (EEPROM), a CD-ROM, a Blu-ray disc, an optical disc storage device, a hard disk drive (HDD), a solid-state drive (SSD), or a flash memory.

[0066] The processor 12 may control the overall functions of the chip 1. For example, the processor 12 may generally control the chip 1 by executing a program stored in the memory 11. The processor 12 may be implemented as a central processing unit (CPU), a graphics processing unit (GPU), an application processor (AP), an ISP (image signal processing), a DPU (display processor), or an NPU (neural processing unit) included in the chip 1 for processing data, but is not limited thereto.

[0067] The processor 12 may read data (eg, image data) from the memory 11 or write data (eg, image data) to the memory 11 and perform purple fringing correction by using the read data / written data.

[0068] In the following, reference will be made to Figures 2 to 5 An example of a hybrid type purple fringing correction method performed by the processor 12 is described.

[0069] The embodiments of the present application provide a mixed type purple fringing correction method, system, electronic device and medium, which are described using the YUV domain as an example, but are not limited to the YUV domain. In the purple fringing detection process, the present application fully considers the formation causes and distribution characteristics of mixed type purple fringing, and uses the mixed dimensional integral of brightness, hue and spatial intensity to make the purple fringing detection intensity have a smooth spatial transition; in the purple fringing correction process, the formation causes and distribution laws of mixed type purple fringing are fully considered, the filter window is adaptively adjusted, and purple fringes of different types and distributions are removed through two steps of filtering and desaturation. Among them, mixed type purple fringing includes purple fringing of highlight overflow type and purple fringing caused by lens chromatic aberration (CA).

[0070] The technical solutions in the embodiments of the present application will be described in detail below in conjunction with the drawings in the embodiments of the present application.

[0071] like Figure 2-5 As shown, this embodiment provides a hybrid type purple fringing correction method, which includes the following steps S1 to S6.

[0072] Step S1, obtaining a color original image.

[0073] Specifically, the color original image obtained in the present application is a YUV image, but is not limited to a YUV image, and may also be an RGB domain image or a RAW domain image, wherein inputY represents the brightness value of the input image, and inputU and inputV represent the chromaticity value of the input image.

[0074] The RGB domain is a color image representation method that generates various colors based on different combinations of the three basic colors of red (R), green (G), and blue (B). The RAW domain refers to the unprocessed data directly output by the image sensor, which contains the most original image signal information.

[0075] Step S2: Calculate the corresponding purple fringing intensity based on the basic information of each pixel in the original image and the basic information of the corresponding first candidate point. The basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located.

[0076] Specifically, in the process of purple fringing detection, the present application takes into account the causes and distribution patterns of purple fringing, and locates the purple fringing and the detection intensity of the purple fringing through two dimensions. On the one hand, local highlight and high contrast are judged based on the brightness value (i.e., Y value) of the pixel points in the original image. On the other hand, the chromaticity range of the pixel points in the original image is judged based on the hue (i.e., U value and V value).

[0077] In one embodiment of the present application, in the purple fringing detection process, calculating the corresponding purple fringing intensity based on the basic information of the pixel points in the original image and the basic information of the corresponding first candidate points includes the following steps S21 to S24.

[0078] Step S21: Calculate the corresponding highlight dimension score based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point.

[0079] Step S22: Calculate the corresponding high contrast dimension score based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point.

[0080] Step S23: Calculate the corresponding hue and saturation score based on the hue of each pixel in the original image.

[0081] Step S24, calculating the corresponding purple fringing intensity according to the highlight dimension score, the high contrast dimension score and the hue saturation score corresponding to each pixel point.

[0082] It should be noted that in the process of purple fringing detection, the process of obtaining the highlight dimension score, the high contrast dimension score and the hue saturation score corresponding to each pixel point is a parallel process, and there is no order of precedence, that is, step S21, step S22 and step S23 are a parallel process.

[0083] For the highlight and high contrast dimensions, first, the highlight surrounding area and the high contrast area are considered. For the purple fringing of the highlight overflow type, the score of this dimension is mainly provided by the highlight point in the local window; for the purple fringing caused by lens chromatic aberration (CA), the score of this dimension is mainly provided by the points with a high brightness difference between the local window and the central pixel point. At the same time, when considering the scores of the brightness and contrast dimensions, both the brightness value itself and the distance from these participating points to the central pixel point are referenced.

[0084] In one embodiment of the present application, the corresponding highlight dimension score is calculated based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point, including the following steps S211 to S214.

[0085] Step S211: Calculate the brightness difference score of each pixel point based on the brightness value of the first candidate point corresponding to each pixel point in the original image and a preset highlight threshold.

[0086] Step S212: Calculate the highlight spatial intensity score of each pixel based on the distance between each pixel and each highlight pixel in the detection area.

[0087] Step S213: Calculate the highlight pixel score of each pixel based on the brightness value of each pixel, the brightness value of the corresponding first candidate point and the preset highlight fluctuation range.

[0088] Step S214: Calculate the corresponding highlight dimension score according to the brightness difference score of each pixel, the highlight spatial intensity score and the highlight pixel score.

[0089] In one embodiment, the judgment and intensity calculation of the "highlight" condition of the present application are as follows: taking the current pixel point as the center, taking M×N Y channel data, polling point by point, setting double thresholds for highlight judgment, a high threshold high_threshold and a low threshold low_threshold, and judging whether the brightness (Y channel data) of each point in the window is higher than or equal to the low threshold low_threshold. If it is lower than the low_threshold, the highlight integration condition is not met and jumps to the next point; if it is higher than or equal to the low_threshold, the highlight integration condition is entered, the brightness of the candidate point is higher than or equal to the high_threshold, and the integral is the highest (for example, the maximum integral is 8), and the brightness of the candidate point is equal to the low_threshold, for example, the integral is 1; the brightness of the candidate point is between the low_threshold and the high_threshold, and the integral number is a linear distance weighted average of 1 and 8. When the brightness of the candidate point is lower than the low_threshold, the highlight integration condition is not met at this time, and the integral can be set to 0 to facilitate the subsequent calculation of the highlight high contrast score, thereby obtaining the brightness difference score of the current pixel point according to the brightness distribution difference of the highlight point. At the same time, the distance from each highlight point to the central pixel is considered, and the second dimension integral is obtained according to the distance. The distance from the point in the window to the central pixel is scaled and limited to [1, 8]. The farther the distance, the lower the score (closer to 1), and the closer the distance, the higher the score (closer to 8). Therefore, according to the spatial distribution difference of the highlight points, the highlight spatial intensity score of the current pixel can be obtained. The integral of the two dimensions of the brightness difference score and the highlight spatial intensity score of the current pixel is added to obtain the total score of the highlight of the M×N window Luma_score_ori. Finally, consider the distribution law of purple fringing, exclude the situation that the central pixel itself is the highlight point, set a fluctuation range high_luma_range of the highlight Y channel data, and perform equal-weight integration on the points whose brightness value is greater than the brightness value Y0+high_luma_range of the central pixel in the M×N window to obtain high_luma_cnt (highlight pixel score), map and limit high_luma_cnt to [1, 8], and adjust Luma_score_ori to obtain the final score of the highlight: Luma_score = (Luma_score_ori×high_luma_cnt) / 8.

[0090] Among them, M×N represents a submatrix centered on the current pixel, and obtains the Y channel data in the submatrix; M represents the height of the submatrix; N represents the width of the submatrix.

[0091] In one embodiment of the present application, the steps of calculating the corresponding high contrast dimension score based on the brightness value of the pixel point in the original image and the brightness value of the corresponding first candidate point include the following steps S221 to S224.

[0092] Step S221: Calculate the brightness difference of each pixel based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate pixel.

[0093] Step S222: Calculate the contrast difference score of each pixel based on the brightness difference of each pixel and a preset contrast threshold.

[0094] Step S223: Calculate the contrast spatial intensity score of each pixel based on the distance between each pixel and each high-contrast pixel in the detection area.

[0095] Step S224: Calculate the corresponding high contrast dimension score according to the contrast difference score and the contrast space intensity score of each pixel.

[0096] In one embodiment, the application judges and calculates the intensity of the "high contrast" condition as follows: taking the current pixel as the center, taking M×N Y channel data, polling point by point, setting a double threshold for high contrast judgment, a high threshold high_contrast_th and a low threshold low_contrast_th, judging whether the difference (Y–Y_center) between the brightness of each point in the window (Y channel data) and the brightness of the current pixel (Y_center) is higher than or equal to the low threshold low_contrast_th. If it is lower than low_contrast_th, the highlight integration condition is not met, and jump to the next point; if it is higher than or equal to low_contrast_th, the highlight integration condition is not met, and jump to the next point; if it is higher than or equal to low_contrast_th, the highlight integration condition is not met, and jump to the next point. h, then enter the highlight integral condition, the brightness of the candidate point is higher than or equal to high_contrast_th, the integral is the highest (for example, the integral is 8), the brightness of the candidate point is equal to low_contrast_th, for example, the integral is 1; the brightness of the candidate point is between low_contrast_th and high_contrast_th, the integral is a linear distance weighted average of 1 and 8; when the brightness of the candidate point is lower than low_contrast_th, the highlight integral condition is not met at this time, and the integral can be set to 0 to facilitate the subsequent calculation of the highlight high contrast score, thereby obtaining the contrast difference score of the current pixel point according to the different degrees of local contrast. At the same time, considering the distance from the point that meets the "high contrast" to the central pixel point, the second dimension integral is obtained according to the distance, and the distance from the point in the window to the central pixel point is scaled and limited to [1,8]. The farther the distance, the lower the score (close to 1), and the closer the distance, the higher the score (close to 8). Therefore, the contrast spatial intensity score of the current pixel point can be obtained according to the spatial distribution difference of the point that meets the "high contrast". The integrals of the two dimensions are added together to obtain the total score High_contrast_score of the high contrast of the M×N window (i.e., the high contrast dimension score of the current pixel).

[0097] The present application obtains the final highlight high contrast score by fusing the highlight dimension score and the high contrast dimension score of the current pixel: Luma_contrast_score = (Luma_score + High_contrast_score) / 2.

[0098] It should be noted that the method of fusing the highlight dimension score and the high contrast dimension score of the current pixel is not limited to the above-mentioned method of averaging the highlight dimension score and the high contrast dimension score of the current pixel, and for example also includes the following fusion method: Luma_contrast_score = (Luma_score×weight_luma+High_contrast_score×weight_contrast) / 8;

[0099] Among them: weight_luma represents the weight of Luma_score; weight_contrast represents the weight of High_contrast_score; 0<=weight_luma<=8; 0<=weight_contrast<=8; weight_luma+weight_contrast=8.

[0100] Note that in addition to the two fusion methods mentioned above, it also includes a weighted sum fusion method, a product fusion method and / or a gradient-based fusion method. The weighted sum fusion method assigns different weights to the highlight dimension score and the high contrast dimension score, respectively, and then multiplies the two and sums them to obtain the fusion result. The product fusion method multiplies the highlight dimension score and the high contrast dimension score to obtain the fusion result. The gradient-based fusion method first calculates the gradient of the pixel point in the highlight dimension and the high contrast dimension, and then fuses the two gradients. This application is not limited to this and will not be described in detail here.

[0101] In the process of purple fringing detection, the present application refers to two factors in the highlight and high contrast dimensions to calculate the detection intensity of purple fringing. On the one hand, it refers to the degree of highlight or high contrast, and on the other hand, it refers to the distance from each pixel to the highlight area. The detected purple fringing intensity has a natural spatial transition, and the image after purple fringing correction can also be smoothly connected with the normal area, avoiding local mutations and unnatural transitions in the processed image.

[0102] In one embodiment of the present application, calculating the corresponding hue and saturation score based on the hue of the pixel points in the original image includes the following steps S231 to S233.

[0103] Step S231: Calculate a corresponding purple score based on the hue of the pixel in the original image, the set purple range and the set purple transition range.

[0104] Step S232: Calculate a corresponding saturation score based on the saturation of the pixel in the original image and a set saturation range.

[0105] Step S233: Calculate the corresponding hue and saturation score according to the purple score and saturation score of each pixel.

[0106] In one embodiment, for the hue and saturation dimensions, considering that the purple fringing is not strictly purple, the purple fringing in outdoor scenes during the day is usually purple, and the "purple fringing" around lights at night is usually bluish purple or even blue. Through the statistical experiments on "purple fringing" on RGB, YUV, and HSV, it is found that the Hue of HSV has stronger robustness for the hue positioning of the purple fringing, and sets the strict "purple" boundaries purple_low_th (the lower boundary of "purple") and purple_high_th (the upper boundary of "purple"), and at the same time, sets the strict "purple" to "non-purple" transition range purple_fade_range. In the hue and saturation score calculation stage, the application first converts the YUV domain into the HSV domain, and then calculates the corresponding hue and saturation score through the Hue value of the central pixel. Among them, HSV is a color space model, also known as the Hexcone Model. The HSV color model describes color through three dimensions: hue, saturation, and value.

[0107] When the Hue of the central pixel is within [purple_low_th, purple_high_th], the chromaticity dimension score is the highest (for example, the chromaticity dimension score is 32 points).

[0108] When the central pixel point Hue is lower than the transition range of the "purple" lower boundary (purple_low_th-purple_fade_range) or higher than the transition range of the "purple" upper boundary (purple_high_th+purple_fade_range), the score is the lowest (for example, the chromaticity dimension score is 0).

[0109] When the center pixel Hue is in the interval [purple_low_th-purple_fade_range, purple_low_th] or [purple_high_th, purple_high_th+purple_fade_range], it is mapped according to the linear distance to obtain a score in the range of [1, 32].

[0110] Thus, the score Purple_score_ori according to the strictness of purple is obtained.

[0111] At the same time, considering the saturation factor, the initial chromaticity score Purple_score_ori is adjusted according to the HSV Saturation. The higher the saturation, the higher the purple fringing intensity, and the lower the saturation, the lower the purple fringing intensity. The saturation range is limited and then linearly mapped to [0, 8] to obtain the saturation score sat_score. The final chromaticity score is obtained by the chromaticity score and the saturation score: Purple_score = (Purple_score_ori × sat_score) / 8.

[0112] The above steps give the "highlight and high contrast" score Luma_score and the "hue saturation" score Purple_score, which are multiplied to get the joint purple fringing score PF_score_ori:

[0113] PF_score_ori=Luma_score×Purple_score;

[0114] Then the joint purple fringing score PF_score_ori is linearly mapped to the range of [0,64] to obtain the final detected purple fringing intensity (PF_score), which has a good spatial transition. At the same time, the detected purple fringing intensity will also serve as a basic reference for the subsequent purple fringing removal intensity.

[0115] In the purple fringing detection stage, the present application is suitable for mixed-type purple fringing removal by considering both highlight conditions and high-contrast conditions. It can remove typical purple fringing caused by overflow of highlight areas, as well as radial purple fringing caused by lens chromatic aberration (CA, Chromatic Aberration). The highlight condition is combined with the color condition, mainly targeting the purple fringing of the highlight overflow type, locating the area where the highlight overflow forms the purple fringing, and correcting it with the information of the adjacent non-highlight area to achieve the purpose of removing the purple fringing; the high-contrast condition is combined with the color condition, mainly targeting the purple fringing of the CA type, taking into account that the position where the chromatic aberration occurs is related to the distribution of pixels at different positions in the image, and the width of the purple fringing formed by the chromatic aberration is also related to the distribution of pixels at different positions in the image, in the correction process, more adjacent pixels in the corresponding direction are used to correct the purple-fringed pixels to achieve the purpose of removing the purple fringing.

[0116] Step S3: determine whether it is a purple-edge pixel according to the purple-edge intensity of each pixel. If it is not a purple-edge pixel, skip the current pixel; if it is a purple-edge pixel, use the set image calibration strategy to obtain the adaptive filtering window of the current purple-edge pixel. The set image calibration strategy is a CAC calibration method, but the present application is not limited to the CAC calibration method.

[0117] like Figure 3As shown, in one embodiment of the present application, determining whether a pixel is a purple-fringed pixel according to the purple-fringing intensity of each pixel includes the following steps S31 to S32.

[0118] S31 . Determine according to the purple fringing intensity of each pixel point. If the purple fringing intensity of the current pixel point is greater than 0, the current pixel point is a purple fringing pixel point.

[0119] S32: If the purple-fringe intensity of the current pixel is equal to 0, the current pixel is not a purple-fringe pixel.

[0120] In an embodiment of the present application, obtaining an adaptive filtering window of a current purple-fringe pixel by setting an image calibration strategy includes the following steps S311 to S312.

[0121] Step S311: Calibrate the original image using a set image calibration strategy to obtain a set of purple-fringe offsets corresponding to purple-fringe pixel points at different positions in the original image.

[0122] Step S312: adaptively generate an adaptive filtering window for the current purple-fringe pixel according to the purple offset set corresponding to the current purple-fringe pixel.

[0123] Specifically, the present application regards pixels with purple-fringe intensity greater than 0 as purple-fringe pixels and constructs corresponding adaptive filtering windows for the purple-fringe pixels, thereby avoiding redundant calculations and improving processing efficiency.

[0124] In the process of correcting purple fringing, the present application targets the spatial distribution laws of chromatic aberration type purple fringing and mixed type purple fringing, and adaptively distributes the size of the correction window and the weights of candidate points in different directions within the correction window participating in the correction. This can make the color of the corrected purple fringing area closer to the original color and make it less likely to obtain low-confidence points for correction. It can better balance the two requirements of "complete purple fringing removal" and "less image error damage".

[0125] Step S4, calculating the candidate compensation color corresponding to each purple-fringed pixel point according to the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringing intensity.

[0126] In one embodiment of the present application, calculating the candidate compensation color corresponding to each purple-fringed pixel point according to the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringing intensity includes the following steps S41 to S44.

[0127] Step S41, acquiring corresponding second candidate pixels according to the adaptive filtering window of each purple-fringe pixel; the second candidate pixels are other pixels within the adaptive filtering window of the current purple-fringe pixel.

[0128] Step S42, calculating the brightness difference of each second candidate pixel corresponding to each purple-fringed pixel according to the brightness value of each purple-fringed pixel and the brightness value of each corresponding second candidate pixel; the brightness difference is the brightness difference between each second candidate pixel and the current purple-fringed pixel.

[0129] Step S43, obtaining the hue saturation score of each second candidate pixel point corresponding to each purple-fringed pixel point according to the hue calculation of each second candidate pixel point corresponding to each purple-fringed pixel point.

[0130] Step S44, calculating the filtered chromaticity of each purple-fringed pixel point according to the brightness difference and chromaticity saturation score of each second candidate pixel point corresponding to each purple-fringed pixel point as a candidate compensation color.

[0131] In one embodiment, in the purple fringing removal stage, the present application performs purple fringing removal processing in two stages: neighborhood compensation color and saturation reduction.

[0132] In one embodiment, the neighborhood compensation color is obtained by finding a point without purple fringes in a local window and using its chromaticity as the compensation color to cover the original "purple". When selecting the compensation color acquisition window, the present application fully considers the formation cause and distribution law of mixed type purple fringes and adaptively acquires the compensation color using the window.

[0133] The CA caused by the physical characteristics of the lens roughly conforms to the quadratic function model. The radiation range of the purple fringing is narrower near the center of the image (usually the image center and the optical center coincide or are very close), and the radiation range of the purple fringing is wider away from the center of the image. At the same time, the radiation direction of the purple fringing usually surrounds the highlight and appears in the direction away from the center of the image.

[0134] Based on the characteristics of mixed-type purple fringing, this application refers to the CA calibration data of the lens to adaptively generate compensation color acquisition windows for each area, namely, adaptive filtering windows. The adaptive filtering window obtains points in the correct direction to the greatest extent to participate in the compensation color filtering, provides more correct points for the compensation color filter, and reduces the problem of cross-color or halo caused by points in the wrong direction. The results of the existing CAC calibration are used on CAC for CA correction of the lens (completed in the RAW domain).

[0135] In one embodiment, the present application utilizes the characteristic that CAC calibration directly affects the interval and width of the purple fringing, and the offset generated by the CAC calibration method is used to generate an adaptive filtering window (a small window will be opened in the center of the image, which is sufficient to cover the central purple fringing while minimizing accidental injury and color smudging; a large window will be adaptively opened at the edge of the image, which is sufficient to cover the purple fringing at the very wide edge).

[0136] In this embodiment, the application obtains the adaptive filtering window based on the calibration or evaluation of CA. First, the image is divided into 16×16 blocks, and the "edge purple offset" of 17×17×2 is obtained by calibration. Among them, the matrix 17×17 corresponds to the value of each vertex of the 16×16 block, and the offset is 2-dimensional, divided into horizontal and vertical directions. The 2-dimensional offset of the 17×17 grid is obtained by calibration and image evaluation.

[0137] Furthermore, in order to reduce the resources of table storage, the matrix 17×17 is divided into 17×2 tables in horizontal and vertical dimensions.

[0138] Window_shift_ori_hor

[17] , used to store the horizontal "purple edge" offset on the edges of 16 blocks;

[0139] Window_shift_ori_ver

[17] , used to store the vertical “purple edge” offset on the edges of 16 blocks;

[0140] Furthermore, the center of the image and the optical center of the lens are normally coincident or very close. The data in the offset storage table can be further reduced to generate two final 9×2 tables.

[0141] Window_shift_hor[9] is used to store the horizontal "purple edge" offset of the edges of the eight blocks on the right half of the image and the "purple edge" offset on the left side of the image. The value at the symmetrical position is taken and then the value is inverted.

[0142] Window_shift_ver[9] is used to store the vertical offset of the "purple fringing" on the edge of the 8 blocks in the lower half of the image and the "purple fringing" offset on the upper side of the image. The value at the symmetrical position is taken and then the value is inverted.

[0143] It should be noted that the size and weight distribution of the adaptive filter window in this application can be directly derived from CA calibration to better fit the purple-fringe spatial distribution characteristics of the lens. However, it is not limited to using CA calibration as a guide, and the image space observation and estimation method can also be used directly to play a similar role.

[0144] In one embodiment, the present application obtains the compensation color by using joint range filtering in the filtering stage. The acquisition of the compensation color is based on two dimensions: the "brightness difference" weightY of the candidate point in the window and the point to be filtered and the "purple edge intensity" weightC of the candidate point in the window (see Figure 5 , the pixel to be processed is the point to be filtered, and the candidate points in the window are other pixel points in the spatially adaptive filtering window).

[0145] Among them, weightY is obtained by subtracting the Y channel data of the candidate point in the window and the Y channel data of the current pixel point and taking the absolute value, and weightC is the hue saturation score obtained by the previous detection step.

[0146] Brightness difference is used as the weight guide for range filtering. The candidate point with brightness closer to the current pixel obtains a larger filtering weight during the filtering process, and the candidate point with brightness farther away from the current pixel obtains a smaller filtering weight during the filtering process.

[0147] The purple fringing intensity is used as the weight guide for range filtering. The candidate point with smaller purple fringing intensity obtains a larger filtering weight during the filtering process, and the candidate point with larger purple fringing intensity obtains a smaller filtering weight during the filtering process.

[0148] In the entire filtering window, U and V of the current pixel are filtered according to the weights of the two dimensions to obtain filtered U and V (ie, depurple_U and depurple_V) as candidate compensation colors for the current pixel.

[0149]

[0150] Where k is the total number of pixels in the current filter window.

[0151] In the compensation color acquisition stage of purple fringing correction, this application uses joint range filtering for compensation color acquisition, and generates filtering weights by referring to two dimensions, namely, the two dimensions of hue and brightness difference. In the hue dimension, smaller weights are given to points with colors close to purple and near purple to ensure that the filtered points are closer to the original colors. In the brightness difference dimension, smaller weights are given to points with large brightness differences to ensure that points with non-homogeneous pixels in the neighborhood are less involved in filtering, thereby reducing the problems of cross-color and color overflow that are more common in range filtering.

[0152] It should be noted that in the compensation color generation stage, the two-dimensional connection and range filtering method used in this application refers to the hue dimension and the brightness difference dimension. However, it is not limited to the hue dimension and the brightness difference dimension. Using the difference of each channel on RGB or the color difference as a guide reference can also play a similar role, which will not be described in detail here.

[0153] In the desaturation stage, in some application scenarios, the purple fringing removal method that simply uses the "compensation color" may not be able to remove the purple fringing completely. For example, in the following two typical application scenarios, in the application scenario where the purple fringing area is too large (close to or exceeds the range of the filter window), the filter candidate points obtained from the local area by the compensation color are also purple fringed, resulting in purple after filtering. In the application scenario where the color saturation of the purple fringing is too high, the points that usually participate in the filtering with a smaller weight may still bring in a certain degree of purple residue due to their high saturation. The original purple fringing area changes from high-saturation purple or blue to light purple or blue after filtering.

[0154] In order to further remove the purple fringing completely, the present application further judges and processes the filtered U and V (i.e., depurple_U and depurple_V). The judgment step is also divided into two dimensions: highlight and high contrast dimension judgment and hue dimension judgment. The judgment conditions of these two dimensions are more stringent to reduce accidental damage caused by desaturation operations.

[0155] In one embodiment, the present application performs highlight and high contrast dimension judgment by adopting the same judgment logic as the highlight and high contrast judgment used in the aforementioned purple fringing detection process, but in this step, the threshold condition is set more strictly.

[0156] Specifically, in this step, there are still high and low thresholds [HL_high_th, HL_low_th] for "highlight" judgment and high and low thresholds [HC_high_th, HC_low_th] for "high contrast". Compared with the high and low thresholds [high_threshold, low_threshold] for "highlight" judgment and high and low thresholds [high_contrast_th, low_contrast_th] for color compensation processing in the previous purple fringing detection step, the following conditions are usually met:

[0157] HL_high_th>high_threshold

[0158] HL_low_th>low_threshold

[0159] HC_high_th>high_contrast_th

[0160] HC_low_th>low_contrast_th

[0161] In this step, the brightness of the highlighted pixels and the weight brought by the distance are retained, so that a dimensional score with a smooth spatial transition can be obtained.

[0162] In one embodiment, the present application performs hue dimension judgment by directly performing a joint judgment of the UV two-dimensional plane on the depurple_U and depurple_V obtained by filtering in the previous step. In this step, in both the U and V directions, the present application sets upper and lower thresholds (U's upper and lower thresholds [U_low_th, U_high_th], V's upper and lower thresholds [V_low_th, V_high_th]) and a transition interval trans_range.

[0163] Specifically, firstly, the highest "de-saturation strength" (ie, high de-saturation strength) is set through the parameter de_sat_strength.

[0164] When the distribution of depurple_U and depurple_V in the UV two-dimensional plane is within the setting range of the above four thresholds, it is regarded as a strict purple-fringe area to be processed, and the highest "de-saturation intensity" is obtained. That is, depurple_U is within [U_low_th, U_high_th], and depurple_V is within [V_low_th, V_high_th].

[0165] Otherwise, when the distribution of depurple_U and depurple_V in the UV two-dimensional plane is within the setting range of the above four thresholds plus the transition interval, it is regarded as the transition purple fringing area to be processed, and a gradually decreasing "de-saturation intensity" is obtained. That is, depurple_U is within [U_low_th-trans_range, U_low_th] or [U_high_th, U_high_th+trans_range], and depurple_V is within [V_low_th-trans_range, V_low_th] or [V_high_th, V_high_th+trans_range].

[0166] When the distribution of depurple_U and depurple_V in the UV two-dimensional plane is outside the setting range of the above four thresholds plus the transition interval, it is considered not a "purple fringing area to be processed" and the desaturation intensity is 0. That is, depurple_U is lower than U_low_th-trans_range or higher than U_high_th+trans_range, and depurple_V is lower than V_low_th-trans_range or higher than V_high_th+trans_range.

[0167] After obtaining the "de-saturation intensity" Dstr of the current pixel, the depurple_U and depurple_V after the above filtering are directly de-saturated to obtain the final UV (ie, new_U and new_V) as the target compensation color of the current pixel.

[0168] new_U=(depurple_U-mid_U)×(1-Dstr)+mid_U

[0169] new_V=(depurple_V-mid_V)×(1-Dstr)+mid_V

[0170] Among them, mid_U and mid_V represent the median of the UV value range, representing the U and V values ​​of mid-grey, corresponding to the UV value of saturation 0.

[0171] It should be noted that in the desaturation stage of the present application, desaturation is directly performed on the UV in the YUV domain, but is not limited to desaturation on the UV. For example, according to the calculated desaturation intensity, processing S on HSV, processing CbCr on CbCr, or processing on the difference between channels of RGB can all play a similar role, and no specific limitation is made here.

[0172] Step S5, calculating the corresponding desaturation intensity according to the candidate compensation color of each purple-fringed pixel, the basic information and the basic information of the corresponding first candidate pixel, and desaturating the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel to obtain the target compensation color corresponding to each purple-fringed pixel. The second candidate pixel is other pixels within the adaptive filtering window of the current purple-fringed pixel.

[0173] In one embodiment of the present application, a corresponding desaturation intensity is calculated based on the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding first candidate pixel point, and the corresponding candidate compensation color is desaturated based on the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point, including the following steps S51 to S53.

[0174] Step S51, calculating a corresponding highlight high contrast score according to the brightness value of each purple-fringe pixel and the brightness value of the corresponding first candidate pixel; the highlight high contrast score is calculated by the highlight dimension score and the high contrast dimension score.

[0175] Step S52, calculating the corresponding desaturation intensity according to the highlight high contrast score of each purple-fringe pixel, the candidate compensation color and the set high desaturation intensity;

[0176] Step S53 : desaturate the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point.

[0177] Figure 4 Shown is a schematic diagram of the overall process of the hybrid type purple fringing correction method described in an embodiment of the present application. Figure 4 The steps S1'-S5' in the embodiment correspond to the above steps S1-S5 respectively, and will not be described in detail here.

[0178] Step S6: Correct the corresponding original pixel points according to the target compensation color of each purple-fringed pixel point to obtain a color image after the purple fringing is removed.

[0179] The present application can better ensure the natural local transition of the image and reduce the local mutation phenomenon caused by purple fringing removal in the previous methods by fully considering the spatial transition of the detection and correction processes in the purple fringing detection stage and the purple fringing correction stage.

[0180] In the judgment conditions of highlight and high contrast, both the value range and the spatial domain are considered. In the value range dimension, the brightness value of the highlight pixel itself or the difference between the bright pixel and the current pixel is considered; in the spatial domain dimension, the spatial distance from the highlight pixel or the bright pixel to the current pixel is considered. The judgment of highlight and high contrast is integrated through the two dimensions to achieve a smooth spatial transition of the detection intensity.

[0181] In the color judgment condition, the "strict purple area" and "transitional purple area" are considered. The closer to the "strict purple area", the higher the color judgment score; the farther away from the "strict purple area", the lower the color judgment score; the point beyond the "transitional purple area", the color judgment score is 0. This further ensures the smooth transition of the space of the detected purple edge intensity.

[0182] In the process of compensating color correction, purple-fringe pixels of different intensities correspond to corrections of different intensities. The detected and corrected pixels and the undetected normal pixels have better spatial transition after processing, reducing the problems of local jumps and unnatural images after processing caused by previous purple-fringe removal methods.

[0183] During the desaturation process, residual purple fringing pixels of different intensities correspond to desaturation processes of different intensities. The desaturated pixels and the pixels that have not entered the desaturation process are fused by gradient weights after processing, ensuring better spatial connection and reducing the problems of local jumps and unnatural images after processing caused by previous desaturation methods for removing purple fringes.

[0184] The present application performs purple fringing correction processing through two stages: compensating color processing and desaturation processing. In the compensating color processing stage, the original color of the pixel contaminated by purple fringing can be restored. In the desaturation processing stage, for the strong purple fringing and ultra-wide purple fringing that cannot be completely covered in the compensating color processing stage, the desaturation method is further adopted to reduce the residual purple fringing.

[0185] In the compensation color processing stage, the joint value range filtering method is used, which takes into account the weight of the color component and the weight of the brightness difference. The weight of the color component makes the filtered color filter out the purple as much as possible while also being able to blend and transition naturally with the surrounding pixels; the brightness difference weight effectively reduces the problems of cross-color and color overflow that may occur during the purple fringing correction process.

[0186] During the desaturation processing stage, the degree of highlights and the spatial attenuation transition of the highlight radiation area are fully considered, and this attenuation transition is used as a guide to further desaturate the residual purple fringes around the highlights, ensuring a natural spatial transition between desaturated and non-desaturated areas.

[0187] The protection scope of the hybrid type purple fringing correction method described in the embodiment of the present application is not limited to the execution order of the steps listed in the present embodiment. All solutions implemented by adding, reducing or replacing steps in the prior art based on the principles of the present application are included in the protection scope of the present application.

[0188] The embodiment of the present application also provides a hybrid type purple fringing correction system, which can implement the hybrid type purple fringing correction method described in the present application. However, the implementation device of the hybrid type purple fringing correction method described in the present application includes but is not limited to the structure of the hybrid type purple fringing correction system listed in the present embodiment. All structural deformations and replacements of the prior art made according to the principles of the present application are included in the protection scope of the present application.

[0189] like Figure 6 As shown, this embodiment provides a hybrid type purple fringing correction system, and the system 3 includes: an acquisition module 31, a purple fringing intensity calculation module 32, an adaptive filtering window construction module 33, a candidate compensation color calculation module 34, a target compensation color calculation module 35 and a correction module 36.

[0190] The acquisition module 31 is configured to acquire a color original image.

[0191] The purple fringing intensity calculation module 32 is configured to calculate the corresponding purple fringing intensity based on the basic information of each pixel in the original image and the basic information of the corresponding first candidate point. The basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located.

[0192] The adaptive filtering window construction module 33 is configured to determine whether the current pixel is a purple-fringed pixel according to the purple-fringing intensity of each pixel. If it is not a purple-fringed pixel, the current pixel is skipped; if it is a purple-fringed pixel, the adaptive filtering window of the current purple-fringed pixel is obtained by using the set image calibration strategy.

[0193] The candidate compensation color calculation module 34 is configured to calculate the candidate compensation color corresponding to each purple-fringed pixel point according to the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringing intensity.

[0194] The target compensation color calculation module 35 is configured to calculate the corresponding desaturation intensity according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and desaturate the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point. The second candidate pixel point is other pixel points within the adaptive filtering window of the current purple-fringed pixel point.

[0195] The correction module 36 is configured to correct the corresponding original pixel points according to the target compensation color of each purple-fringed pixel point to obtain a color image after the purple fringing is removed.

[0196] It should be noted that the functions or operations of the acquisition module 31, purple fringing intensity calculation module 32, adaptive filtering window construction module 33, candidate compensation color calculation module 34, target compensation color calculation module 35 and correction module 36 described in the embodiment of the present disclosure correspond one by one to the steps in the mixed type purple fringing correction method described above, so they will not be repeated here.

[0197] In the several embodiments provided in the present application, it should be understood that the disclosed system, device or method can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules / units is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or units can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or modules or units, which can be electrical, mechanical or other forms.

[0198] The modules / units described as separate components may or may not be physically separated, and the components displayed as modules / units may or may not be physical modules, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules / units may be selected according to actual needs to achieve the purpose of the embodiments of the present application. For example, the functional modules / units in the various embodiments of the present application may be integrated into one processing module, or each module / unit may exist physically separately, or two or more modules / units may be integrated into one module / unit.

[0199] Those of ordinary skill in the art should further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0200] The embodiment of the present application also provides a chip, the chip includes: a memory and a processor; wherein the memory is used to store computer programs; the memory includes: ROM, RAM, disk, USB flash drive, memory card or optical disk and other media that can store program codes. The processor is used to execute the computer program stored in the memory, so that the electronic device performs the hybrid type purple fringing correction method as described above.

[0201] Preferably, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0202] like Figure 7As shown, the electronic device 4 of the present application is in the form of a general computing device. The components of the control terminal may include but are not limited to: one or more processors or processing units 41, a memory 42, and a bus 43 connecting different system components (including the memory 42 and the processing unit 41).

[0203] Bus 43 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus architectures. By way of example, these architectures include, but are not limited to, an Industry Standard Architecture (ISA) bus, a Micro Channel Architecture (MAC) bus, an Enhanced ISA bus, a Video Electronics Standards Association (VESA) local bus, and a Peripheral Component Interconnect (PCI) bus.

[0204] The control terminal typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the control terminal, including volatile and non-volatile media, removable and non-removable media.

[0205] The memory 42 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 421 and / or cache memory 422. The control terminal may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory 42 may be used to read and write non-removable, non-volatile magnetic media ( Figure 7 not shown, usually called a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to the bus 43 via one or more data medium interfaces. The memory 42 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of the various embodiments of the present disclosure.

[0206] A program / utility 424 having a set (at least one) of program modules 4241 may be stored, for example, in the memory 42, such program modules 4241 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 4241 generally perform the functions and / or methods of the embodiments described in the present disclosure.

[0207] The present application embodiment also provides a computer-readable storage medium. A person of ordinary skill in the art can understand that all or part of the steps in the method for implementing the above embodiment can be completed by a program to instruct the processor, and the program can be stored in a computer-readable storage medium, and the storage medium is a non-transitory medium, such as a random access memory, a read-only memory, a flash memory, a hard disk, a solid-state hard disk, a magnetic tape, a floppy disk, an optical disc, and any combination thereof. The above storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a digital video disc (DVD)), or a semiconductor medium (e.g., a solid-state hard disk (SSD)), etc.

[0208] The present application embodiment may also provide a computer program product, the computer program product including one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the process or function described in the embodiment of the present application is generated in whole or in part. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website site, computer or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer or data center.

[0209] When the computer program product is executed by a computer, the computer executes the method described in the above method embodiment. The computer program product may be a software installation package, and when the above method is required, the computer program product may be downloaded and executed on a computer.

[0210] In summary, the hybrid type purple fringing correction method, system, electronic device and medium described in the present application have the following beneficial effects:

[0211] The present application calculates the corresponding purple fringing intensity based on the basic information of each pixel in the original image and the basic information of the corresponding first candidate point; the basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; according to the purple fringing intensity of each pixel point, the adaptive filtering window of each purple fringing pixel point is obtained by setting the image calibration strategy, and the corresponding candidate compensation color is calculated according to the adaptive filtering window of each purple fringing pixel point and the corresponding purple fringing intensity; the corresponding desaturation intensity is calculated according to the candidate compensation color of each purple fringing pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple fringing pixel point to obtain the target compensation color corresponding to each purple fringing pixel point; the second candidate pixel point is other pixel points in the adaptive filtering window of the current purple fringing pixel point; the corresponding original pixel point is corrected according to the target compensation color of each purple fringing pixel point to obtain a color image after removing the purple fringing. By fully considering the spatial transition of the detection and correction process in the purple fringing detection stage and the purple fringing correction stage, it is possible to better ensure the natural local transition of the image and reduce the local mutation phenomenon caused by removing the purple fringing in the previous method.

[0212] In the judgment conditions of highlight and high contrast, the present application considers both the value range and the spatial domain. In the value range dimension, the brightness value of the highlight pixel itself, or the difference between the bright pixel and the current pixel, is considered; in the spatial domain dimension, the spatial distance from the highlight pixel or the bright pixel to the current pixel is considered. The judgment of highlight and high contrast is integrated in two dimensions to achieve a smooth spatial transition of the detection intensity. In the judgment conditions of color, the "strict purple area" and the "transitional purple area" are considered. The closer to the "strict purple area", the higher the color discrimination score; the farther away from the "strict purple area", the lower the color discrimination score; the point beyond the "transitional purple area", the color discrimination score is 0. It can further ensure the smooth spatial transition of the detected purple edge intensity.

[0213] In the process of compensating color correction in the present application, purple-fringed pixels of different intensities correspond to corrections of different intensities. The detected and corrected pixels and the normal pixels that are not detected have better spatial transition after processing, reducing the local jumps and unnatural images caused by previous purple-fringing removal methods. At the same time, in the process of desaturation processing, residual purple-fringed pixels of different intensities correspond to desaturation processing of different intensities. The desaturated pixels and the pixels that have not entered the desaturation processing are fused by gradient weights after processing, ensuring better spatial connection and reducing the local jumps and unnatural images caused by previous desaturation purple-fringing removal methods.

[0214] The present application performs purple fringing correction processing through two stages: compensating color processing and desaturation processing. In the compensating color processing stage, the original color of the pixel contaminated by purple fringing can be restored. In the desaturation processing, for the strong purple fringing and ultra-wide purple fringing that cannot be completely covered in the compensating color processing stage, the desaturation method is further adopted to reduce the residual purple fringing. In the compensating color processing stage, the joint value range filtering method is used, and the weights of the color components and the brightness difference weights are considered at the same time. The weights of the color components can filter out the purple as much as possible after filtering, and can also be naturally blended and transitioned with the surrounding pixels; the brightness difference weights can effectively reduce the problems of cross-color and color overflow that may occur during the purple fringing correction process. In the desaturation processing stage, the degree of highlight and the spatial attenuation transition of the highlight radiation area are fully considered, and the attenuation transition is used as a guide to further desaturate the residual purple fringing around the highlight, ensuring the natural spatial transition between the desaturated area and the non-desaturated area.

[0215] The descriptions of the processes or structures corresponding to the above-mentioned figures have different emphases. For parts that are not described in detail in a certain process or structure, please refer to the relevant descriptions of other processes or structures.

[0216] The above embodiments are merely illustrative of the principles and effects of the present application and are not intended to limit the present application. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present application. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed in the present application shall still be covered by the claims of the present application.

Claims

1. A hybrid type purple fringing correction method, characterized in that: The method comprises: Get the original color image; The corresponding purple fringing intensity is calculated based on the basic information of each pixel in the original image and the basic information of the corresponding first candidate point; the basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; Determine whether it is a purple-fringed pixel according to the purple-fringing intensity of each pixel. If it is not a purple-fringed pixel, skip the current pixel. If it is a purple-fringed pixel, use the set image calibration strategy to obtain the adaptive filtering window of the current purple-fringed pixel. The candidate compensation color corresponding to each purple-fringed pixel point is obtained by calculating the adaptive filtering window of each purple-fringed pixel point and the corresponding purple-fringed intensity; The corresponding desaturation intensity is calculated according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point; the second candidate pixel point is other pixel points within the adaptive filtering window of the current purple-fringed pixel point; The corresponding original pixel is corrected according to the target compensation color of each purple-fringed pixel to obtain a color image after the purple-fringed pixel is removed.

2. The hybrid type purple fringing correction method according to claim 1, characterized in that: Determining whether the current pixel is a purple-fringed pixel according to the purple-fringing intensity of each pixel includes: The purple fringing intensity of each pixel is used to determine if the purple fringing intensity of the current pixel is greater than 0, and the current pixel is a purple fringed pixel. If the purple fringing intensity of the current pixel is equal to 0, the current pixel is not a purple fringed pixel.

3. The hybrid type purple fringing correction method according to claim 1, characterized in that: The adaptive filtering window of the current purple-edge pixel point is obtained by setting the image calibration strategy, including: Calibrate the original image by using a set image calibration strategy to obtain a set of purple-fringe offsets corresponding to purple-fringe pixel points at different positions in the original image; An adaptive filtering window for the current purple-fringe pixel is adaptively generated according to the purple offset set corresponding to the current purple-fringe pixel.

4. The hybrid type purple fringing correction method according to claim 1, characterized in that: The candidate compensation colors corresponding to each purple-fringed pixel are calculated based on the adaptive filtering window of each purple-fringed pixel and the corresponding purple-fringed intensity, including: Acquire corresponding second candidate pixels according to the adaptive filtering window of each purple-edge pixel; the second candidate pixels are other pixels within the adaptive filtering window of the current purple-edge pixel; The brightness difference of each second candidate pixel corresponding to each purple-fringed pixel is calculated according to the brightness value of each purple-fringed pixel and the brightness value of each corresponding second candidate pixel; the brightness difference is the brightness difference between each second candidate pixel and the current purple-fringed pixel; The chroma saturation score of each second candidate pixel point corresponding to each purple-fringed pixel point is obtained according to the chroma of each second candidate pixel point corresponding to each purple-fringed pixel point; The chromaticity of each purple-fringed pixel after filtering is calculated based on the brightness difference and chromaticity saturation score of each second candidate pixel corresponding to each purple-fringed pixel as a candidate compensation color.

5. The hybrid type purple fringing correction method according to claim 1, characterized in that: The corresponding desaturation intensity is calculated according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and the corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple-fringed pixel point to obtain the target compensation color corresponding to each purple-fringed pixel point, including: The corresponding highlight high contrast score is calculated according to the brightness value of each purple-fringe pixel and the brightness value of the corresponding second candidate pixel; the highlight high contrast score is calculated by the highlight dimension score and the high contrast dimension score; The corresponding desaturation intensity is calculated according to the highlight high contrast score of each purple-fringe pixel, the candidate compensation color and the set high desaturation intensity; The corresponding candidate compensation color is desaturated according to the desaturation intensity of each purple-fringe pixel point to obtain the target compensation color corresponding to each purple-fringe pixel point.

6. The hybrid type purple fringing correction method according to claim 1, characterized in that: In the purple fringing detection process, the corresponding purple fringing intensity is calculated based on the basic information of each pixel point in the original image and the basic information of the corresponding first candidate point, including: Calculate the corresponding highlight dimension score based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point; Calculate the corresponding high contrast dimension score based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point; Calculate the corresponding hue and saturation score of each pixel in the original image based on the hue; The corresponding purple fringing intensity is calculated according to the highlight dimension score, the high contrast dimension score and the hue saturation score corresponding to each pixel point.

7. The hybrid type purple fringing correction method according to claim 6, characterized in that: The corresponding highlight dimension score is calculated based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate point, including: The brightness difference score of each pixel point is calculated based on the brightness value of the first candidate point corresponding to each pixel point in the original image and a preset highlight threshold; The highlight spatial intensity score of each pixel is calculated based on the distance between each pixel and each highlight pixel in the detection area; The highlight pixel score of each pixel is calculated based on the brightness value of each pixel, the brightness value of the corresponding first candidate point and the preset highlight fluctuation range; The corresponding highlight dimension score is calculated based on the brightness difference score of each pixel, the highlight spatial intensity score and the highlight pixel score.

8. The hybrid type purple fringing correction method according to claim 6, characterized in that: The corresponding high contrast dimension score is calculated based on the brightness value of the pixel point in the original image and the brightness value of the corresponding first candidate point, including: Calculate the brightness difference of each pixel based on the brightness value of each pixel in the original image and the brightness value of the corresponding first candidate pixel; The contrast difference score of each pixel is calculated based on the brightness difference of each pixel and the preset contrast threshold; The contrast spatial intensity score of each pixel is calculated based on the distance between each pixel and each high-contrast pixel in the detection area; The corresponding high contrast dimension score is calculated based on the contrast difference score and contrast spatial intensity score of each pixel.

9. The hybrid type purple fringing correction method according to claim 6, characterized in that: The corresponding hue saturation score calculated based on the hue of the pixel in the original image includes: Calculate a corresponding purple score based on the hue of the pixel in the original image, the set purple range and the set purple transition range; Calculate the corresponding saturation score based on the saturation of the pixel in the original image and the set saturation range; The corresponding hue-saturation score is calculated based on the purple score and saturation score of each pixel.

10. A hybrid type purple fringing correction system, characterized in that: The system comprises: An acquisition module is configured to acquire a color original image; A purple fringing intensity calculation module is configured to calculate the corresponding purple fringing intensity based on basic information of each pixel in the original image and basic information of the corresponding first candidate point; the basic information includes brightness value and hue; the first candidate point is other pixel points in the detection area where the current pixel point is located; The adaptive filtering window construction module is configured to determine whether the current pixel is a purple-fringed pixel according to the purple-fringing intensity of each pixel, and if it is not a purple-fringed pixel, skip the current pixel; if it is a purple-fringed pixel, obtain the adaptive filtering window of the current purple-fringed pixel by using the set image calibration strategy; A candidate compensation color calculation module is configured to calculate a candidate compensation color corresponding to each purple-fringed pixel point according to an adaptive filtering window of each purple-fringed pixel point and a corresponding purple-fringed intensity; a target compensation color calculation module, configured to calculate a corresponding desaturation intensity according to the candidate compensation color of each purple-fringed pixel point, the basic information and the basic information of the corresponding second candidate pixel point, and desaturate the corresponding candidate compensation color according to the desaturation intensity of each purple-fringed pixel point to obtain a target compensation color corresponding to each purple-fringed pixel point; the second candidate pixel point is other pixel points within the adaptive filtering window of the current purple-fringed pixel point; The correction module is configured to correct the corresponding original pixel points according to the target compensation color of each purple-fringed pixel point to obtain a color image after the purple fringing is removed.

11. An electronic device, characterized in that: The electronic device comprises: a memory and a processor; The memory is configured to store a computer program; The processor is configured to execute the computer program stored in the memory, so as to enable the electronic device to perform the hybrid type purple fringing correction method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by an electronic device, the hybrid type purple fringing correction method according to any one of claims 1 to 9 is implemented.

Images