Extremely low illumination tobacco image enhancement method based on adaptive parameter optimization

Abstract

The invention discloses an ultra-low illumination tobacco image enhancement method based on adaptive parameter optimization, and belongs to the technical field of tobacco image enhancement. The method comprises the following steps: S1, acquiring extremely low illumination tobacco leaf image data; s2, splitting the colored ultra-low illumination tobacco leaf image into R, G and B color channels; s3, dynamically adjusting a color channel through an adaptive linear distribution enhancement method, and improving the image contrast and the noise suppression capability; s4, recombining the dynamically adjusted color channels according to an original sequence to generate a recombined image; s5, processing the recombined image by using a detail enhancement algorithm, and recovering textures and details on the basis of retaining the natural color of the tobacco leaves to obtain a detail enhanced image; and S6, performing contrast stretching or brightness correction on the detail enhanced image by using a contrast adjustment algorithm, and finally obtaining a curing barn tobacco leaf curing enhanced image. According to the method, the problems of insufficient image brightness, fuzzy details, noise interference, low contrast ratio, color distortion and the like in the curing barn environment are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of tobacco leaf image enhancement, and in particular to an extremely low-light tobacco leaf image enhancement method based on adaptive parameter optimization. Background Art

[0002] With the widespread application of low-light image enhancement technology in the tobacco industry, application scenarios have been applied to nighttime tobacco leaf monitoring, agricultural monitoring in rainy weather, cigar drying monitoring, and tobacco leaf baking in flue-curing barns. In recent years, breakthroughs in zero-sample and few-sample low-light image enhancement technology have provided new possibilities for extremely low-light image enhancement in flue-curing barns under data-scarce conditions. However, there are still many technical bottlenecks in the current ultra-low-light image enhancement in flue-curing barns: loss of details and blurring during the enhancement process, insufficient image contrast, noise amplification caused by brightness enhancement, and color distortion. Although deep learning methods have shown significant advantages in the field of image enhancement, their dependence on high-quality training data and the difficulty in obtaining samples in the actual flue-curing barn environment restrict their practical application in tobacco leaf baking monitoring. Summary of the invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and provide an extremely low-light tobacco leaf image enhancement method based on adaptive parameter optimization.

[0004] The object of the present invention is achieved through the following technical solutions: The first aspect of the present invention provides: a method for enhancing an extremely low-light tobacco leaf image based on adaptive parameter optimization, comprising the following steps: S1: Acquire extremely low light tobacco leaf image data; S2: Split the color extremely low-light tobacco leaf image into three color channels: R, G, and B; S3: Dynamically adjust the color channel through the adaptive linear distribution enhancement method to improve image contrast and noise suppression capabilities; S4: recombining the dynamically adjusted color channels in the original order to generate a recombined image; S5: Use detail enhancement algorithm to process the reconstructed image, restore the texture and details on the basis of retaining the natural color of tobacco leaves to obtain a detail enhanced image; S6: Use a contrast adjustment algorithm to perform contrast stretching or brightness correction on the detail enhanced image, and finally obtain a flue-curing tobacco leaf baking enhanced image.

[0005] Preferably, the adaptive linear distribution enhancement method comprises the following steps: Perform image normalization, and then calculate image statistical features and histograms; after obtaining image statistical features, determine whether its average pixel value is less than 100. If it is less than 100, set the gamma coefficient γ to 0.7 to generate the target distribution; if it is greater than or equal to 100, set the gamma coefficient γ to 1 to generate the target distribution; after obtaining the histogram, calculate the cumulative distribution; after obtaining the target distribution and cumulative distribution, use the mapping function for interpolation mapping; finally, perform local histogram equalization; The average pixel value calculation formula is: μ = (1 / MN)∑∑I(i,j), where μ represents the average pixel value of the target image or image area; M represents the number of rows of the target image or image area, and N represents the number of columns of the target image or image area; i represents the row coordinate of the pixel, j represents the column coordinate of the pixel, and I(i,j) represents the pixel value at the coordinate (i,j); The histogram calculation formula is: h(k) =∑∑δ(I(i,j)-k), k∈[0,255], where h(k) represents the number of pixels with gray value k in the target image or image area; δ() represents the counting function; k represents the gray level; The cumulative distribution function is: , where CDF(k) represents the proportion of pixels with grayscale values ​​less than or equal to k to the total pixels; The target distribution function is: , where T(x) represents the output pixel value; x represents the input pixel value; γ represents the gamma coefficient; The mapping function is: M(x)=interp(CDF(k),T(x),[0,255]); The local histogram equalization formula is: H local (x,y) = clip(h local (x,y) / N local , limit), where H local (x,y) represents the normalized histogram after clipping; h local (x,y) represents the original histogram of the local area; N local represents the total number of pixels in the local area; limit = 3.0 × M1N1 / 256 represents the clipping threshold, M1 represents the number of rows in the local window, and N1 represents the number of columns in the local window; clip() represents the clipping function.

[0006] Preferably, the detail enhancement algorithm comprises the following steps: Convert the reconstructed image from RGB to LAB color space, separate the LAB channel, and process the L channel; Perform bilateral filtering, then extract the detail layer, and then use the gain coefficient to enhance the image; Finally, merge the base layer and detail layer, set the limit value range, and merge the LAB channels to obtain a detail-enhanced image; The bilateral filtering formula is: ,in I base ( x ) represents the pixel value at position x after bilateral filtering; Wp represents the normalization factor; I ( y ) represents the original image pixel value at position y; Gaussian weights representing spatial distances; Gaussian weights representing differences in pixel values; Represents a neighborhood centered on a pixel; The detail layer extraction formula is: detail (x) = I(x) - I base ( x ), where I detail (x) represents the extracted detail component; I(x) represents the original input image; The gain factor application formula is: I detail_enhanced (x) = gain · I detail (x), where I detail_enhanced (x) represents the image enhanced by the gain factor; gain represents the gain factor; Detail enhanced image L enhanced (x) = clip( I base ( x ) + I detail_enhanced (x), 0, 255).

[0007] Preferably, the contrast adjustment algorithm comprises the following steps: Check the input range, find the maximum and minimum values ​​of the input image, and then calculate the cumulative distribution; Then the scaling factor is calculated and the first normalization is performed; Perform boundary processing on the image after the first normalization processing to determine whether its output value is less than 0. If so, set it to 255. If it is greater than or equal to 0, determine whether it is greater than 255. If so, set it to 255. If it is less than or equal to 255 and greater than or equal to 0, keep the original value. Normalize the image after boundary processing again: ,in I final Represents the pixel value of the image after final conversion; round () means converting a floating point number to the nearest integer; I clipped Represents the input value after image boundary processing.

[0008] The second aspect of the present invention provides: an extremely low-light tobacco leaf image enhancement system based on multi-scale fusion, which is used to implement any of the extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization, comprising: An image acquisition module, used to acquire extremely low-light tobacco leaf image data; An image splitting module is used to split the color extremely low-light tobacco leaf image into three color channels: R, G, and B; The image enhancement module is used to dynamically adjust the color channels through an adaptive linear distribution enhancement method to improve the image contrast and noise suppression capabilities; recombine the dynamically adjusted color channels in the original order to generate a reconstructed image; use a detail enhancement algorithm to process the reconstructed image, restore the texture and details on the basis of retaining the natural color of the tobacco leaves to obtain a detail enhanced image; use a contrast adjustment algorithm to perform contrast stretching or brightness correction on the detail enhanced image, and finally obtain a flue-curing tobacco leaf baking enhanced image.

[0009] The third aspect of the present invention provides: a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization is implemented.

[0010] A fourth aspect of the present invention provides: a computer program product comprising instructions, which, when executed on a terminal, enables the terminal to execute any of the above-mentioned extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization.

[0011] The beneficial effects of the present invention are: 1) By integrating the improved adaptive linear distribution enhancement, adaptive detail enhancement and contrast adjustment algorithms, the problems of insufficient image brightness, blurred details, noise interference, low contrast and color distortion in the flue-curing room environment are effectively solved. Even under the condition of severe image degradation, high-quality tobacco leaf baking images can be provided, providing a reliable data basis for the automated monitoring of tobacco leaf baking. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The flowchart of the extremely low-light tobacco leaf image enhancement method based on adaptive parameter optimization is shown; Figure 2 Enhance the algorithm flow chart for details; Figure 3 Flowchart for the contrast adjustment algorithm. DETAILED DESCRIPTION

[0013] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0014] See also Figure 1-Figure 3 The first aspect of the present invention provides: a method for enhancing an extremely low-light tobacco leaf image based on adaptive parameter optimization, comprising the following steps: S1: Acquire extremely low light tobacco leaf image data; S2: Split the color extremely low-light tobacco leaf image into three color channels: R, G, and B; S3: Dynamically adjust the color channel through the adaptive linear distribution enhancement method to improve image contrast and noise suppression capabilities; S4: recombining the dynamically adjusted color channels in the original order to generate a recombined image; S5: Use detail enhancement algorithm to process the reconstructed image, restore the texture and details on the basis of retaining the natural color of tobacco leaves to obtain a detail enhanced image; S6: Use a contrast adjustment algorithm to perform contrast stretching or brightness correction on the detail enhanced image, and finally obtain a flue-curing tobacco leaf baking enhanced image.

[0015] In this embodiment, the zero-sample adaptive extremely low-light image enhancement technology is used to break through the application bottleneck of traditional enhancement methods in the complex environment of the flue-curing barn. This method comprehensively improves the image quality on the basis of ensuring efficient calculation, effectively suppresses noise interference, maintains the integrity of details, and achieves accurate color restoration. The image enhancement effect of this method under extremely low light conditions is significantly better than that of the existing technology, especially in an environment such as a flue-curing barn with complex and changeable lighting conditions. It can enhance the image of tobacco leaves under extremely low light conditions, overcoming the defects of the existing low-light tobacco leaf image enhancement technology.

[0016] In some embodiments, the adaptive linear distribution enhancement method comprises the following steps: Perform image normalization, and then calculate image statistical features and histograms; after obtaining image statistical features, determine whether its average pixel value is less than 100. If it is less than 100, set the gamma coefficient γ to 0.7 to generate the target distribution; if it is greater than or equal to 100, set the gamma coefficient γ to 1 to generate the target distribution; after obtaining the histogram, calculate the cumulative distribution; after obtaining the target distribution and cumulative distribution, use the mapping function for interpolation mapping; finally, perform local histogram equalization; The average pixel value calculation formula is: μ = (1 / MN)∑∑I(i,j), where μ represents the average pixel value of the target image or image area; M represents the number of rows of the target image or image area, and N represents the number of columns of the target image or image area; i represents the row coordinate of the pixel, j represents the column coordinate of the pixel, and I(i,j) represents the pixel value at the coordinate (i,j); The histogram calculation formula is: h(k) =∑∑δ(I(i,j)-k), k∈[0,255], where h(k) represents the number of pixels with gray value k in the target image or image area; δ() represents the counting function; k represents the gray level; The cumulative distribution function is: , where CDF(k) represents the proportion of pixels with grayscale values ​​less than or equal to k to the total pixels; The target distribution function is: , where T(x) represents the output pixel value; x represents the input pixel value; γ represents the gamma coefficient; The mapping function is: M(x)=interp(CDF(k),T(x),[0,255]); The local histogram equalization formula is: H local (x,y) = clip(h local (x,y) / N local , limit), where H local (x,y) represents the normalized histogram after clipping; h local (x,y) represents the original histogram of the local area; N local represents the total number of pixels in the local area; limit = 3.0 × M1N1 / 256 represents the clipping threshold, M1 represents the number of rows in the local window, and N1 represents the number of columns in the local window; clip() represents the clipping function.

[0017] In some embodiments, the detail enhancement algorithm comprises the following steps: Convert the reconstructed image from RGB to LAB color space, separate the LAB channel, and process the L channel; Perform bilateral filtering, then extract the detail layer, and then use the gain coefficient to enhance the image; Finally, merge the base layer and detail layer, set the limit value range, and merge the LAB channels to obtain a detail-enhanced image; The bilateral filtering formula is: ,in I base ( x ) represents the pixel value at position x after bilateral filtering; Wp represents the normalization factor; I ( y) represents the original image pixel value at position y; Gaussian weights representing spatial distances; Gaussian weights representing differences in pixel values; Represents a neighborhood centered on a pixel; The detail layer extraction formula is: detail (x) = I(x) - I base ( x ), where I detail (x) represents the extracted detail component; I(x) represents the original input image; The gain factor application formula is: I detail_enhanced (x) = gain · I detail (x), where I detail_enhanced (x) represents the image enhanced by the gain factor; gain represents the gain factor; Detail enhanced image L enhanced (x) = clip( I base ( x ) + I detail_enhanced (x), 0, 255).

[0018] In some embodiments, the contrast adjustment algorithm comprises the following steps: Check the input range, find the maximum and minimum values ​​of the input image, and then calculate the cumulative distribution; Then the scaling factor is calculated and the first normalization is performed; Perform boundary processing on the image after the first normalization processing to determine whether its output value is less than 0. If so, set it to 255. If it is greater than or equal to 0, determine whether it is greater than 255. If so, set it to 255. If it is less than or equal to 255 and greater than or equal to 0, keep the original value. Normalize the image after boundary processing again: ,in I final Represents the pixel value of the image after final conversion; round () means converting a floating point number to the nearest integer; I clipped Represents the input value after image boundary processing.

[0019] The second aspect of the present invention provides: an extremely low-light tobacco leaf image enhancement system based on multi-scale fusion, which is used to implement any of the extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization, comprising: An image acquisition module, used to acquire extremely low-light tobacco leaf image data; An image splitting module is used to split the color extremely low-light tobacco leaf image into three color channels: R, G, and B; The image enhancement module is used to dynamically adjust the color channels through an adaptive linear distribution enhancement method to improve the image contrast and noise suppression capabilities; recombine the dynamically adjusted color channels in the original order to generate a reconstructed image; use a detail enhancement algorithm to process the reconstructed image, restore the texture and details on the basis of retaining the natural color of the tobacco leaves to obtain a detail enhanced image; use a contrast adjustment algorithm to perform contrast stretching or brightness correction on the detail enhanced image, and finally obtain a flue-curing tobacco leaf baking enhanced image.

[0020] The third aspect of the present invention provides: a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by a processor, any of the above-mentioned extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization is implemented.

[0021] A fourth aspect of the present invention provides: a computer program product comprising instructions, which, when executed on a terminal, enables the terminal to execute any of the above-mentioned extremely low-light tobacco leaf image enhancement methods based on adaptive parameter optimization.

[0022] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A method for enhancing tobacco leaf images in extremely low light conditions based on adaptive parameter optimization, characterized in that: The following steps are involved: S1: Acquire extremely low light tobacco leaf image data; S2: Split the color extremely low-light tobacco leaf image into three color channels: R, G, and B; S3: Dynamically adjust the color channel through the adaptive linear distribution enhancement method to improve image contrast and noise suppression capabilities; S4: recombining the dynamically adjusted color channels in the original order to generate a recombined image; S5: Use detail enhancement algorithm to process the reconstructed image, restore the texture and details on the basis of retaining the natural color of tobacco leaves to obtain a detail enhanced image; S6: Use a contrast adjustment algorithm to perform contrast stretching or brightness correction on the detail enhanced image, and finally obtain a flue-curing tobacco leaf baking enhanced image.

2. The method for extremely low-light tobacco leaf image enhancement based on adaptive parameter optimization according to claim 1, characterized in that: The adaptive linear distribution enhancement method comprises the following steps: Perform image normalization, and then calculate image statistical features and histograms; after obtaining image statistical features, determine whether its average pixel value is less than 100. If it is less than 100, set the gamma coefficient γ to 0.7 to generate the target distribution; if it is greater than or equal to 100, set the gamma coefficient γ to 1 to generate the target distribution; after obtaining the histogram, calculate the cumulative distribution; after obtaining the target distribution and cumulative distribution, use the mapping function for interpolation mapping; finally, perform local histogram equalization; The average pixel value calculation formula is: μ=(1 / MN)∑∑I(i,j), where μ represents the average pixel value of the target image or image area; M represents the number of rows of the target image or image area, and N represents the number of columns of the target image or image area; i represents the row coordinate of the pixel, j represents the column coordinate of the pixel, and I(i,j) represents the pixel value at the coordinate (i,j); The histogram calculation formula is: h(k) =∑∑δ(I(i,j)-k), k∈[0,255], where h(k) represents the number of pixels with gray value k in the target image or image area; δ() represents the counting function; k represents the gray level; The cumulative distribution function is: , where CDF(k) represents the proportion of pixels with grayscale values ​​less than or equal to k to the total pixels; The target distribution function is: , where T(x) represents the output pixel value; x represents the input pixel value; γ represents the gamma coefficient; The mapping function is: M(x)=interp(CDF(k),T(x),[0,255]); The local histogram equalization formula is: H_local(x,y) = clip(h_local(x,y) / N_local, limit), where H_local(x,y) represents the normalized histogram after clipping; h_local(x,y) represents the original histogram of the local area; N_local represents the total number of pixels in the local area; limit =3.0 × M1N1 / 256 represents the clipping threshold, M1 represents the number of rows of the local window, and N1 represents the number of columns of the local window; clip() represents the clipping function.

3. The method for enhancing tobacco leaf images under extremely low illumination based on adaptive parameter optimization according to claim 1, characterized in that: The detail enhancement algorithm comprises the following steps: Convert the reconstructed image from RGB to LAB color space, separate the LAB channel, and process the L channel; Perform bilateral filtering, then extract the detail layer, and then use the gain coefficient to enhance the image; Finally, merge the base layer and detail layer, set the limit value range, and merge the LAB channels to obtain a detail-enhanced image; The bilateral filtering formula is: ,in I base ( x ) represents the pixel value at position x after bilateral filtering; Wp represents the normalization factor; I ( y ) represents the original image pixel value at position y; Gaussian weights representing spatial distances; Gaussian weights representing differences in pixel values; Represents a neighborhood centered on a pixel; The detail layer extraction formula is: detail (x) = I(x) - I base ( x ), where I detail (x) represents the extracted detail component; I(x) represents the original input image; The gain factor application formula is: I detail_enhanced (x) = gain · I detail (x), where I detail_enhanced (x) represents the image enhanced by the gain factor; gain represents the gain factor; Detail enhanced image L enhanced (x) = clip( I base ( x ) + I detail_enhanced (x), 0, 255).

4. The method for extremely low-light tobacco leaf image enhancement based on adaptive parameter optimization according to claim 1, characterized in that: The contrast adjustment algorithm comprises the following steps: Check the input range, find the maximum and minimum values ​​of the input image, and then calculate the cumulative distribution; Then the scaling factor is calculated and the first normalization is performed; Perform boundary processing on the image after the first normalization processing to determine whether its output value is less than 0. If so, set it to 255. If it is greater than or equal to 0, determine whether it is greater than 255. If so, set it to 255. If it is less than or equal to 255 and greater than or equal to 0, keep the original value. Normalize the image after boundary processing again: ,in I final Represents the pixel value of the image after final conversion; round () means converting a floating point number to the nearest integer; I clipped Represents the input value after image boundary processing.

5. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the computer-executable instructions are loaded and executed by the processor, the extremely low-light tobacco leaf image enhancement method based on adaptive parameter optimization as described in any one of claims 1 to 4 is implemented.

6. A computer program product comprising instructions, characterized in that: When the computer program product is run on a terminal, the terminal executes the extremely low-light tobacco leaf image enhancement method based on adaptive parameter optimization as described in any one of claims 1 to 4.

Images