An adaptive enhancement method for non-uniform illumination images

Abstract

This invention discloses an adaptive enhancement method for images with non-uniform illumination. First, for an input image in RGB color space, its Y channel in YUV color space is calculated and low-pass filtered to serve as the initial brightness channel. Second, a pixel-level threshold is calculated based on the local brightness of the image to determine the exposure level of each pixel. Then, using this threshold as the center, a symmetrical brightness remapping function is used to adaptively adjust the local brightness, thereby eliminating overexposure and underexposure. Finally, the adjusted brightness channel is combined with the original image to reconstruct the RGB channels of the color image and optimize the local contrast to obtain the enhanced result. This adaptive enhancement method can achieve effective enhancement results for images with complex illumination variations, especially those with small areas of extreme illumination.

Description

Technical Field

[0001] The invention relates to an adaptive enhancement method for non-uniform illumination images, belonging to the technical field of image processing. Background Art

[0002] When there are significant differences in ambient lighting, objects in the scene receive uneven illumination, resulting in regional variations in brightness in the resulting color image. On the one hand, areas with stronger lighting appear overexposed, severely compressing details and colors; on the other hand, areas with weaker lighting appear underexposed, drowning out details and colors.

[0003] Therefore, enhancing images with non-uniform illumination requires correcting both bright and dark areas. Traditional single brightness adjustment methods, such as gamma transform and histogram equalization, are ineffective and can easily lead to over- or under-enhancement of small areas. Therefore, it is necessary to design an adaptive enhancement scheme that leverages local information to boost the brightness of underexposed areas and suppress the brightness of overexposed areas.

[0004] Patent CN 113112429 A decomposes an image into a base layer and a detail layer, using the base layer's grayscale mean as a threshold. Regions with grayscale values ​​greater than this threshold are considered overexposed and their brightness should be suppressed, while the brightness of the remaining regions should be boosted. However, for images with large areas of overexposure or underexposure, the grayscale mean is relatively extreme, which can easily lead to underenhancement of small areas within the image. Patent CN 110992287 A estimates image brightness through a series of maximum filtering and uses the median of the image's grayscale range as an invisible threshold to distinguish between highlights and dark areas, remapping the image brightness. However, a single empirical threshold cannot cope with complex lighting variations. Recognizing the limitations of a single threshold, Patent CN 114331873 A uses a series of binarization operations to obtain image boundaries, thereby dividing the image into multiple regions. The exposure level of each region is then determined to adaptively adjust the brightness of each region. However, this method relies heavily on the accuracy of region segmentation. In addition to the heuristic brightness adjustment function, the existing technology has studied non-uniform illumination image enhancement methods based on deep learning. Patent CN 115526803 A proposes an unsupervised enhancement method based on a differential neural network model and an entropy maximization loss function, but its enhancement effect on highlight areas in the image is not significant.

[0005] For images with non-uniform illumination, the heuristic single threshold used to define highlight and dark areas often cannot adapt to the complex changes in image content and illumination, and easily causes under-enhancement in some areas, affecting the enhancement effect and versatility. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide an adaptive enhancement method for images with non-uniform illumination. This method proposes a locally adaptive threshold for determining the exposure level of local pixels, thereby avoiding the local under-enhancement caused by a single threshold. Furthermore, a symmetrical brightness mapping function is designed based on this threshold to adaptively control the adjustment range of local brightness, thereby reducing local over-enhancement.

[0007] The present invention adopts the following technical solutions to solve the above technical problems:

[0008] An adaptive enhancement method for non-uniform illumination images comprises the following steps:

[0009] Step 1: For the original non-uniform illumination color image, the grayscale values ​​of its RGB channels are used to calculate the Y channel value in the YUV space to obtain the brightness channel of the original non-uniform illumination color image;

[0010] Step 2: Use an anisotropic bilateral filter to perform a convolution operation on the brightness channel obtained in step 1 to estimate the illumination intensity map of the original non-uniform illumination color image;

[0011] Step 3: Calculate an adaptive threshold map for determining exposure levels based on the light intensity map obtained in step 2, and determine the exposure levels of each pixel in the original non-uniform illumination color image;

[0012] Step 4: remap the brightness channel of the original non-uniform illumination color image using a brightness remapping function according to the exposure level of each pixel to obtain a remapped brightness map;

[0013] Step 5, reconstructing the original non-uniform illumination color image according to the remapped brightness image obtained in step 4 to obtain a reconstructed color image;

[0014] Step 6: Use an exponential function to enhance the local contrast of the reconstructed color image obtained in step 5 to obtain the final adaptive enhancement result image.

[0015] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0016] 1. For images with non-uniform illumination, this paper designs an adaptive threshold to determine the exposure level of a local area based on the local illumination intensity of the image. Compared to heuristic thresholds, adaptive thresholds can more flexibly determine underexposure and overexposure in small areas. It is suitable for images with complex ambient lighting variations and can effectively avoid under-enhancement in small areas.

[0017] 2. The present invention combines the designed adaptive threshold with the brightness remapping function to effectively control the brightness adjustment amplitude of the local area, thereby avoiding the result distortion caused by local over-enhancement.

[0018] 3. The adaptive enhancement method designed in this invention can achieve clear images by taking into account both highlights and dark areas in images with drastic lighting changes. This method can achieve effective enhancement results for images with complex lighting variations, especially those with small areas of extreme lighting. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the adaptive enhancement method for non-uniform illumination images proposed in the present invention;

[0020] Figure 2 is a color image and its threshold map Example;

[0021] Figure 3 Is a specific parameter and threshold Schematic diagram of the brightness remapping curve below;

[0022] Figure 4 It is an indoor non-uniform illumination image and its enhanced result, where (a) is the presence of indoor backlight, (b) is the indoor single light source, and (c) is the indoor single light source;

[0023] Figure 5 It is an outdoor non-uniform illumination image and its enhanced result;

[0024] Figure 6 Here is an underexposed image and its enhanced result. DETAILED DESCRIPTION

[0025] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be interpreted as limiting the present invention.

[0026] like Figure 1 As shown, for non-uniform illumination images, the present invention proposes an adaptive enhancement method, the specific process is as follows:

[0027] (1) Input a non-uniform illumination color image and obtain its grayscale images of each channel in the RGB color space, which are recorded as .

[0028] (2) Calculate the brightness channel of the image, that is, use the RGB channel grayscale image to calculate the Y channel value of the YUV space. The specific formula is as follows:

[0029] ,

[0030] in, and represents the horizontal and vertical coordinates of the pixels in the image, Pixels The brightness value, Represents pixels exist Grayscale value of the channel.

[0031] (3) In the Y channel, a convolution operation is performed using an anisotropic bilateral filter to estimate the image's illumination intensity map. , the specific formula is:

[0032] ,

[0033] in, Pixels The estimated lighting, Indicated in pixels The circular neighborhood area with centered on is , and the radius of the area is one fifth of the diagonal length of the original color image. Pixels Grayscale value in the Y channel. For bilateral filter at pixel and The weight value between is calculated as follows:

[0034] ,

[0035] Among them, the spatial domain variance The value is one tenth of the diagonal length of the original color image, and the grayscale domain variance The value is one tenth of the difference between the maximum grayscale and the minimum grayscale in the luminance channel Y.

[0036] (4) Image illumination map calculated based on (3) , calculate the adaptive threshold map used to judge the exposure level , used to determine the exposure level of each pixel.

[0037] Generally speaking, pixels with higher light intensity should be judged as more severely overexposed, so as to guide the subsequent steps to suppress their brightness and restore the details; conversely, pixels with lower light intensity should be judged as more severely underexposed, so as to guide the subsequent steps to increase their brightness. Designing an Adaptive Threshold Map , the specific formula is:

[0038] ,

[0039] Among them, the parameters To control the threshold The maximum value of .parameter To characterize the threshold The rate of change with brightness. The recommended value is 1.5. Figure 2 Shown is the threshold map of the example image , whose value changes inversely with the pixel illumination. Based on the threshold map , for any pixel , if the estimated illumination Greater than the corresponding threshold , it is judged as overexposed; otherwise, it is judged as underexposed.

[0040] (5) Threshold map calculated based on (4) , we can determine the exposure level of each pixel. According to the exposure level, the brightness channel Y of the input image is remapped to correct the exposure problem in the image. Specifically, at the threshold At , the monotonically increasing Naka-Rushton function and its symmetric function about the grayscale median (0.5, 0.5) are spliced ​​to obtain a brightness remapping function that can simultaneously improve underexposure and suppress overexposure, as follows:

[0041] ,

[0042] in, From (2), we can get: It can be calculated by (4). Adjustable parameters Used to control the amplitude of brightness remapping, the value is the mean of the brightness channel of the input image The functions are as follows:

[0043] ,

[0044] Where, It is an absolute value operation. Specifically, the brightness remapping function is and threshold The change curve of Figure 3 shown.

[0045] Depend on Figure 3 It can be seen that for the parameters , when the input pixel brightness Less than threshold , it is judged to be underexposed, and the brightness mapping Greater than its input brightness ; On the contrary, if the input pixel brightness Greater than threshold , it is judged to be overexposed, and the brightness mapping Less than its input brightness In addition, to illustrate the parameters For the effect of brightness remapping, Figure 3 Shows the When fixed, the remap function curve follows According to the parameters The calculation formula is: when the input image brightness mean The closer it is to the median of the image grayscale dynamic range of 0.5, The closer the value is to 1, the closer the brightness remapping function curve is to , that is, the smaller the brightness adjustment, the better the remapping result. The closer it is to the input brightness On the contrary, when the mean brightness of the input image deviates from 0.5, The smaller the value, the greater the adjustment of the brightness remapping.

[0046] (6) Remapped brightness map calculated according to (5) , combined with the input color image , reconstruct the color image after brightness adjustment :

[0047] ,

[0048] Among them, the subscript Represents a channel in the RGB color space. It is an adjustable parameter used to reduce the distortion of color reconstruction caused by brightness changes. The details are as follows:

[0049] ,

[0050] Where, It is an absolute value operation. For control The recommended value is 1. Used to control the effect of brightness change on the color reconstruction of the output image. The recommended value is 0.6. Greater than hour, The value is greater than 1 and It changes positively, which can enhance the color of the highlight area in the input image; when the input grayscale Less than hour, The value is less than 1 and It changes in the opposite direction, which can effectively avoid over-enhancement of colors in dark areas caused by increased brightness.

[0051] (7) For the color image reconstructed in (6) , respectively, in the RGB channel, with the help of exponential function, enhance the local contrast to improve the local details of the image, as follows:

[0052] ,

[0053] Among them, the index part For The result of small-scale Gaussian filtering is calculated as follows:

[0054] ,

[0055] in, Calculated from (6), is the Gaussian filter weight, and the calculation formula is as follows:

[0056] ,

[0057] Where, Represents the Gaussian variance, and the recommended value is one twentieth of the diagonal length of the original color image.

[0058] Figure 4 Shown is a set of indoor non-uniform illumination images and their enhancement results. Figure 4 The problem with (a) is that the indoor backlight causes the indoor objects to be submerged in darkness. Figure 4 The outdoor scene in (a) is clear in details, so it is necessary to maintain the clarity and color of the outdoor scene while enhancing the indoor scene. Figure 4 (b) and (c) are images taken indoors under a single light source. The part close to the light source is overexposed, while the rest of the image is severely underexposed. It is necessary to increase the brightness of the dark areas while suppressing the brightness of the highlight areas.

[0059] Figure 5 The following figure shows an outdoor image with non-uniform illumination and its enhancement results. Outdoor unidirectional sunlight causes shadow areas in the image, so it is necessary to improve the quality of the shadow areas while optimizing the details of the areas under strong light.

[0060] In addition, to demonstrate the versatility of the present invention, Figure 6 An underexposed image and its enhancement result are shown. The present invention can also effectively enhance image details and avoid over-enhancement of colors.

[0061] comprehensive Figure 4-Figure 6 The present invention can adaptively enhance the details of each local area of ​​an image with uneven exposure, avoiding over-enhancement or under-enhancement of a small area, and the obtained result has clear details and balanced colors.

[0062] Based on the same inventive concept, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the aforementioned method for adaptive enhancement of non-uniformly illuminated images are implemented.

[0063] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the aforementioned adaptive enhancement method for non-uniform illumination images are implemented.

[0064] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0066] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0068] The above embodiments are only for illustrating the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.

Claims

1. An adaptive enhancement method for non-uniform illumination images, characterized in that: The steps include: Step 1: For the original non-uniform illumination color image, the grayscale values ​​of its RGB channels are used to calculate the Y channel value in the YUV space to obtain the brightness channel of the original non-uniform illumination color image; Step 2: Use an anisotropic bilateral filter to perform a convolution operation on the brightness channel obtained in step 1 to estimate the illumination intensity map of the original non-uniform illumination color image; Step 3: Calculate an adaptive threshold map for determining exposure levels based on the light intensity map obtained in step 2, and determine the exposure levels of each pixel in the original non-uniform illumination color image; The calculation formula of the adaptive threshold map is as follows: T(x,y)=k(1-(Y BF (x,y)) γ ) Among them, Y BF (x, y) is the estimated illumination of pixel (x, y) in the original non-uniform illumination color image, T(x, y) is the threshold corresponding to pixel (x, y), x and y represent the horizontal and vertical coordinates of pixel (x, y), k is the parameter used to control the maximum value of the threshold, and γ is the parameter used to characterize the rate of change of the threshold with brightness; If the estimated illumination Y of pixel (x,y) BF If (x,y) is greater than the corresponding threshold T(x,y), the pixel (x,y) is judged to be overexposed; otherwise, it is judged to be underexposed; Step 4: remap the brightness channel of the original non-uniform illumination color image using a brightness remapping function according to the exposure level of each pixel to obtain a remapped brightness map; The remapping calculation formula is as follows: Among them, Y(x,y) is the grayscale value of the pixel (x,y) in the Y channel in the original non-uniform illumination color image, and Y map (x,y) is the brightness value of the pixel (x,y) after remapping, and H is an adjustable parameter used to control the brightness remapping amplitude. Its values ​​are as follows: H=1-|Y mean -0.5| Among them, Y mean is the mean grayscale value of all pixels in the original non-uniform illumination color image in the Y channel, |·| is the absolute value operation; Step 5, reconstructing the original non-uniform illumination color image according to the remapped brightness image obtained in step 4 to obtain a reconstructed color image; Step 6: Use an exponential function to enhance the local contrast of the reconstructed color image obtained in step 5 to obtain the final adaptive enhancement result image.

2. The adaptive enhancement method for non-uniform illumination images according to claim 1, characterized in that: In step 1, the Y channel value of the original non-uniform illumination color image in the YUV space is calculated as follows: Y(x,y)=0.299I R (x,y)+0.587I G (x,y)+0.114I B (x,y) Among them, I C (x,y) represents the grayscale value of pixel (x,y) in channel C, C∈{R,G,B}.

3. The adaptive enhancement method for non-uniform illumination images according to claim 1, characterized in that: In step 2, the calculation formula of the light intensity map is as follows: Among them, Y BF (x, y) is the estimated illumination of pixel (x, y) in the original non-uniform illumination color image, Ω(x, y) represents the neighborhood area centered on pixel (x, y), (a, b) represents a pixel in the neighborhood area Ω(x, y), and Y(a, b) is the grayscale value of pixel (a, b) in the Y channel; BF(x, y, a, b) is the weight value of the bilateral filter between pixels (x, y) and (a, b), and is calculated as follows: Among them, σ S is the spatial domain variance, which is one tenth of the diagonal length of the original non-uniform illumination color image; σ R is the grayscale domain variance, which is one tenth of the difference between the maximum grayscale and the minimum grayscale in the Y channel.

4. The adaptive enhancement method for non-uniform illumination images according to claim 1, characterized in that: In step 5, the reconstructed color image is represented as follows: Among them, I C (x, y) represents the grayscale value of the pixel (x, y) in the original non-uniform illumination color image in the C channel, E C (x,y) represents the grayscale value of the pixel (x,y) in the C channel after reconstruction, C∈{R,G,B}; α is an adjustable parameter used to reduce the distortion caused by brightness changes on color reconstruction, as follows: α=w-(λ-I C (x,y))|λ-I C (x,y)| Among them, w is a parameter that controls the value range of α, λ is a parameter used to control the impact of brightness changes on the color reconstruction of the output image, and |·| is an absolute value operation.

5. The adaptive enhancement method for non-uniform illumination images according to claim 1, characterized in that: In step 6, the local contrast of the reconstructed color image is enhanced using an exponential function, and the calculation formula is as follows: Among them, O C (x,y) represents the reconstructed color image E C The grayscale value of the pixel (x, y) in the C channel after contrast enhancement, E C (x,y) represents the grayscale value of the pixel (x,y) in the C channel after reconstruction, GE C (x,y) represents a Gaussian filter, specifically: Among them, Ω(x,y) represents the neighborhood area centered on the pixel (x,y), (i,j) represents a pixel in the neighborhood area Ω(x,y), and E C (i, j) represents the grayscale value of pixel (i, j) in channel C after reconstruction, G(x, y, i, j) is the Gaussian filter weight, and the calculation formula is as follows: Among them, δ G Represents the Gaussian variance, which is one twentieth of the diagonal length of the original non-uniform illumination color image.

6. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the computer program, the steps of the adaptive enhancement method for non-uniform illumination images according to any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for adaptively enhancing a non-uniformly illuminated image according to any one of claims 1 to 5 are implemented.

Images