CN116193276A - An Improved 3A Image Preprocessing Algorithm Based on DM8127 - Google Patents

Abstract

The invention belongs to the technical field of image processing, specifically, it is an improved 3A image preprocessing algorithm based on DM8127. Through the research on the autofocus algorithm, the sharpness evaluation function is analyzed and compared, and the search for the optimal value is improved. way to reduce the number of searches and achieve better search results. Combining the effect of aperture on depth of field blur and the effect of gain on noise, based on the quality evaluation function, a new automatic exposure algorithm is proposed to achieve the best balance of comprehensive image quality between depth of field blur and noise. In the automatic white balance part, the image is partitioned, the gray area is found through the YUV component, and the experiment is carried out on the DM8127 platform, and the parameters are adjusted to make the result converge, which is better than the commonly used automatic white balance algorithm.

Description

An Improved 3A Image Preprocessing Algorithm Based on DM8127

technical field

The invention belongs to the technical field of image processing, in particular, it is an improved 3A image preprocessing algorithm based on DM8127.

Background technique

Nowadays, video is closely related to people's daily life. Under the background of wireless network, high-definition video, high-definition TV, and high-definition calls all reflect the importance of high resolution. If the video resolution obtained by users is low, which affects the viewing experience and the transmission of information, then the value of this video will be reduced, and it will be discarded in the endless network space like garbage. Every day, the same piece of video data undergoes countless round-trip transmissions. People need videos to reach their user devices quickly and obtain the latest information in real time in order to continue to move forward in this competitive society. Of course, the timeliness of video requires that the transmission speed must be fast. Therefore, how to transmit a large amount of important high-definition video information in an open wireless environment, how to improve the quality of the acquired video image, and realize fast and secure transmission has become a hot spot in the field of video processing.

The currently commonly used image preprocessing techniques include color interpolation, color correction, gamma correction, image enhancement, and white balance. The 3A algorithm is to analyze the image through auto-focus, auto-exposure, and auto-white balance to achieve maximum image contrast, improve the overexposure or underexposure of the target subject, and compensate the chromatic aberration of the picture under different light irradiation, so as to present a brighter image. High-quality image information.

The DM8127 chip is specially optimized for security monitoring. It has a unique low-light technology and high-efficiency compression function. It outputs video at a full frame rate of 1080P60 frames/s. It integrates 3D noise reduction, wide dynamic range and strong light suppression processing technology, and face detection. , video stabilization, zoom distortion correction and other image signal processing technologies, in addition, it also adopts the image acquisition technology that supports up to 16 million pixels, combined with the unique 750MHz DSP for intelligent analysis, and is used to realize the high-definition monitoring of the intelligent front end.

Contents of the invention

The present invention takes image preprocessing as the basic starting point, and proposes an improved 3A image preprocessing algorithm based on DM8127. The DM8127 chip has the advantages of high resolution, rich algorithm library, multi-core communication, and superior DSP computing performance. Therefore, this method uses the DM8127 chip as the computing platform, based on the 3A algorithm, to preprocess the image at the receiving end, and to optimize the process of finding the optimal value. Optimized to find the best balance between depth of field blur and noise, improving image quality.

The concrete technical scheme that the present invention adopts is as follows:

An improved 3A image preprocessing algorithm based on DM8127. The video image is sent to the PC through the DM8127 video processing system for image acquisition. The acquisition process is based on the 3A algorithm. The 3A algorithm analyzes the image through auto focus, auto exposure, and auto white balance. , realize the maximum image contrast, improve the overexposure or underexposure of the target subject, and compensate the chromatic aberration of the picture under different light irradiation, thereby presenting higher quality image information.

In the above technical solution, DM8127 is mainly composed of ARM core and DSP core. Among them, the ARM core is a Cortext-A8 processor, and the DSP core is a digital signal processor. It also integrates M3-VPSS core and M3-VIDEO core. Two ARM Cortex-M3 cores are used as coprocessors for the control and management of the High-Definition Video Processing Subsystem (HDVPSS) and the High-Definition Video Image Coprocessor (HDVICP2) respectively. The cores of DM8127 are connected by the L3 internal bus, and the mutual access between the components is realized through the L3 bus. Among them, the ARM core runs the Linux system and calls the VPSS core, Video core, and DSP core to control the realization of the functions of each module. After the M3-VPSS core runs the BIOS system, it realizes image acquisition, display and zooming, etc.; after the M3-VIDEO core runs the BIOS system, it receives the pre-processed image data and is responsible for video image encoding and decoding processing. The DSP core is responsible for the application of the algorithm, and runs the BIOS system to perform image encryption operations. Finally, the ARM core receives the data stream and transmits it over the network.

The further improvement of the present invention optimizes the optimal value of the algorithm, optimizes the global search, divides the global search range into different regions, each region adopts a different step size, and can reduce the number of searches for the focus search problem. Better search results.

In the above technical solution, the automatic exposure process is divided into two steps: automatic gain and automatic aperture. The automatic gain is to cooperate with the aperture to adjust the amount of incoming light. The H3A module of the DM8127 platform calculates the current brightness value curY, sets the target brightness as targetY, and calculates the ratio ratio. The calculation formula is as follows:

Then adjust the three parameters of exposure time, sensor gain and chip digital gain according to the ratio ratio.

The automatic aperture is to calculate the quality evaluation function, change the aperture to find the maximum value, and define the image quality evaluation function F(I) for the aperture as follows:

Among them, the collected image I(x, y) is basically the same as the scene image R(x, y), and the quality of the scene image R(x, y) is unknown, so the quality of the collected image I(x, y) is used as The standard for optimum aperture. h(x, y, σ) is a blur function, and the noise is affected by the gain g, and the best image quality should be the balance between the depth of field blur and the noise.

Automatic white balance includes lighting estimation and image color correction. The image is divided into areas to find the gray area, the white balance algorithm is implemented on the experimental platform, and the parameters are adjusted to make the result converge, which achieves better results than the commonly used automatic white balance algorithm. Automatic white balance includes lighting estimation and image color correction. Grayscale world and perfect reflection are two classic common assumption algorithms. The core idea of the grayscale world is to control the average values of the three channels to be equal, and the core idea of perfect reflection is to control the maximum values of the three channels to be equal. Both methods have their own advantages and disadvantages, and are applicable to different scenarios in different degrees. In order to overcome the defects of the two methods, the present invention proposes an improved algorithm in the step of illumination estimation: the image is partitioned, the closer to gray, the gray world assumption can be satisfied, and the gains of the three channels in each area are adjusted so that The mean of each area is close to gray. Define the image I(, y, y), and add coefficients μ _r , μ _b , γ _r , γ _b to correct the pixels of the image. According to formula (1) and formula (2), new pixel values can be calculated. According to formula (3) and (4) calculate the new pixel mean value, and calculate the new maximum value according to formula (5) and formula (6):

meanG＝μ _r meanR ² +γ _r meanR (3)

meanG＝μ _b meanB ² +γ _b meanB (4)

max(R _new )=max{I _g (x,y)} (5)

max(B _new )=max{I _g (x, y)} (6)

This improved method does not correct the green channel, and satisfies the gray-scale world algorithm, and the mean values of the new red and blue channels are equal to the green channel; satisfies the perfect reflection algorithm, and the maximum values of the new red and blue channels are equal to the green channel, and the formula ( 1) and formula (2) are substituted into formula (5) and formula (6) to get:

According to formula (3), formula (4), formula (7), formula (8), the mean value and maximum value of the red and blue channels are expressed in matrix form, as shown in formula (9) and formula (10):

Using the Gaussian elimination method to solve, the gain coefficients of the red and blue channels are obtained as follows:

This method is tested on the DM8127 platform, and the parameters are adjusted to make the result converge, which is better than the commonly used automatic white balance algorithm.

Beneficial effects of the present invention: the present invention adjusts the value of each channel as close to 0 as possible under the condition of no light; the automatic exposure module adjusts parameters such as exposure time and aperture size, and selects the appropriate light intensity; autofocus Determine the appropriate focal length; automatic white balance is to correct the influence of color according to different scenes; defective pixels in the image are processed in the dead point correction process; 2D noise reduction can reduce noise and improve the accuracy of interpolation; Bayer interpolation produces red, green and blue Three-channel data; gamma correction and color correction are used to correct the contrast and color of the image; the brightness contrast enhancement and edge enhancement modules can improve brightness and enhance image clarity; finally, the distortion correction module corrects the distorted image. Each module of image preprocessing is relatively independent and cooperates with each other to obtain high-quality images.

Description of drawings

FIG. 1 is a flow chart of preprocessing in an embodiment of the present invention.

Fig. 2 is an algorithm flow chart in the embodiment of the present invention.

Fig. 3 is a flow chart of automatic exposure in the embodiment of the present invention.

Fig. 4 is a block diagram of the H3A module in the embodiment of the present invention.

Detailed ways

In order to deepen the understanding of the present invention, the present invention will be described in further detail below in conjunction with the accompanying drawings and embodiments, which are only used to explain the present invention and do not limit the protection scope of the present invention.

Embodiment: as shown in Figure 1, a kind of improved 3A image preprocessing algorithm based on DM8127, send video image to PC through DM8127 video processing system, consider each factor that affects image quality, from image color, illumination intensity, camera focal length and other modules to conduct research, so that each module can adjust and cooperate with each other: under the condition of no light, the value of each channel is as close to 0 as possible through black level adjustment; the automatic exposure module adjusts parameters such as exposure time and aperture size, and selects the appropriate The intensity of light; Auto focus to determine the appropriate focal length; Auto white balance corrects the influence of color according to different scenes; Defective pixels in the image are processed in the dead point correction process; 2D noise reduction can reduce noise and improve the accuracy of interpolation; Bayer interpolation produces red, green and blue three-channel data; gamma correction and color correction are used to correct the contrast and color of the image; brightness contrast enhancement and edge enhancement modules can improve brightness and enhance image clarity; finally, distorted images are corrected in the distortion correction module Make corrections. Each module of image preprocessing is relatively independent and cooperates with each other to obtain high-quality images.

As shown in Figure 2, on the basis of the DM8127 video processing system, the present invention optimizes, studies and adjusts the H3A module to optimize the image quality, and performs image preprocessing through the three hardware modules of ISIF, H3A and IPIPE. At the same time, the preprocessing process also adds 2A (automatic exposure and automatic white balance), scene adaptive local dynamic range enhancement, video stabilization, lens distortion correction, noise filtering, screen software display and other algorithms for optimization and adjustment.

Fig. 3 is a flow chart of automatic exposure in the embodiment of the present invention. First set the initial aperture, and calculate the brightness according to the actual scene. If within the threshold range, the parameters are unchanged. If it exceeds the threshold, judge whether the low-light threshold is met. When it is less than the low-light threshold, select the maximum aperture and maximum exposure time for gain; if it is greater than the low-light threshold, conduct the experiment according to normal light. Under normal lighting, adjust the aperture size according to the quality evaluation function, find the maximum value of the function, the automatic exposure process ends, if the maximum value is not found, repeat the following steps until the maximum value is found: adjust the aperture direction according to the brightness of the scene. If the light intensity becomes larger, then close the adjustment aperture; if the light intensity becomes smaller, then open the adjustment aperture.

Under normal lighting conditions, first adjust the gain, use the H3A module of the DM8127 platform to calculate, and then according to the set target brightness targetY, the calculation formula is as follows:

Then calculate the quality evaluation function of the current image, as follows:

Find the maximum value according to the quality evaluation function, set the aperture to the maximum value, and the automatic exposure process ends; if there is no maximum value, adjust the gain according to the direction of light. If the scene becomes brighter, the aperture should be closed to reduce the depth of field blur; if the scene is darker, the aperture should be opened to reduce the impact of noise, and this process is repeated until the optimal value is found.

As shown in Figure 4, the hardware 3A module of DM8127 provides hardware support for pixel statistics for automatic exposure, and the acquisition process is as follows:

First, the image is down-sampled, and each frame of image is divided into windows, and each window is divided into 2×2 blocks, and each pixel of each block is counted separately, and then the image is saturated. Check, if the pixel of the block exceeds the limit , the number of unsaturated blocks will not be counted, and the count will be counted after replacing the limit value. Finally, each pixel window is accumulated and output.

Through the in-depth research and analysis of the H3A module, the present invention adds the blur caused by the depth of field into the consideration factors of the aperture adjustment through the research on the blur of the depth of field, and proposes an evaluation function for the optimal aperture. Combined with this evaluation function , you can find the optimal aperture size, find the optimal balance point between depth of field blur and noise, and optimize the comprehensive image quality.

The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the industry should understand that the present invention is not limited by the above-mentioned embodiments. What are described in the above-mentioned embodiments and the description only illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will also have Variations and improvements are possible, which fall within the scope of the claimed invention. The protection scope of the present invention is defined by the appended claims and their equivalents.

Claims (6)

1. The improved 3A image preprocessing algorithm based on the DM8127 is characterized in that a DM8127 video processing system sends video images to a PC for image acquisition, the acquisition process is based on a 3A algorithm, and the 3A algorithm specifically comprises automatic focusing, automatic exposure and automatic white balance.

2. The improved 3A image preprocessing algorithm based on DM8127 as set forth in claim 1, wherein the DM8127 VIDEO processing system is composed of an ARM core and a DSP core, wherein the ARM core is a Cortex-A8 processor, the DSP core is a digital signal processor, and two ARM Cortex-M3 cores, an M3-VPSS core and an M3-VIDEO core, are integrated as coprocessors for controlling and managing the high definition VIDEO processing subsystem and the high definition VIDEO image coprocessor, respectively; the cores of the DM8127 are connected by an L3 internal bus, and mutual access among the components is realized through the L3 bus.

3. The improved DM8127 based 3A image preprocessing algorithm as recited in claim 2, wherein the process of automatic exposure is divided into two steps: the automatic gain and the automatic aperture, the target brightness is set as targetY, the ratio is calculated, and the calculation formula is as follows:

and adjusting three parameters of exposure time, sensor gain and chip digital gain according to the ratio.

4. The improved DM8127 based 3A image preprocessing algorithm as set forth in claim 3, wherein the automatic aperture is a calculated quality-evaluation function, the aperture-finding maximum is changed, and the image quality-evaluation function F (I) for the aperture is defined as follows:

wherein the acquired image I (x, y) is substantially identical to the scene image R (x, y) and the scene image R (x, y) is of unknown quality, so the quality of the acquired image I (x, y) is used as a standard of an optimal aperture, h (x, y, sigma) is a blurring function, noise is influenced by the gain g, and the optimal image quality should be a balance of depth blurring and noise.

5. The improved 3A image preprocessing algorithm based on DM8127 as set forth in claim 4, wherein in said automatic exposure process, the image is partitioned, the closer to gray the more can be satisfied the gray world assumption, the gains of three channels in each region are adjusted so that the average value of each region is close to gray, the image I (x, y) is defined, and the new coefficient μ is added _r 、μ _b 、γ _r 、γ _b Pixels for correcting an image, new pixel values are calculated according to the formula (1) and the formula (2), new pixel mean values are calculated according to the formula (3) and the formula (4), and new maximum values are calculated according to the formula (5) and the formula (6):

meanG＝μ _r meanR ² +γ _r meanR (3)

meanG＝μ _b meanB ² +γ _b mmeanB (4)

max(R _new )＝max{I _g (x，y)} (5)

max(B _new )＝max{I _g (x，y)} (6)，

substituting the formula (1) and the formula (2) into the formula (5) and the formula (6) to obtain:

the average value and the maximum value of the red and blue channels are expressed in matrix form according to the formula (3), the formula (4), the formula (7) and the formula (8), as shown in the formulas (9) and (10):

solving by using a Gaussian elimination method to obtain gain coefficients of the red and blue channels as follows:

6. the DM8127 based improved 3A image preprocessing algorithm as recited in claim 5, wherein the DM8127 video processing system 3A module provides hardware support for pixel statistics for automatic exposure, and the acquisition process is as follows: firstly, downsampling an image, dividing each frame of image into windows, dividing each window into 2 multiplied by 2 blocks, respectively counting each pixel of each block, then carrying out saturation examination on the image, if the pixels of the block exceed the limit, counting after replacing the limit value, and finally accumulating and outputting each pixel window without counting the number of unsaturated blocks.

Images