Self-adaptive wide-gray-scale-range infrared image highlight target suppression method

Abstract

The invention relates to a self-adaptive wide-gray-scale-range infrared image highlight target suppression method, and belongs to the technical field of infrared image processing. According to the method, histogram information of an original image is utilized, whether a highlight suppression algorithm processing flow is entered or not is adaptively judged according to a statistical result and a decision function, the histogram of the image is collected and compressed in the flow, and meanwhile, a nonlinear mapping method is used, so that the purpose of effectively suppressing a highlight target is achieved; and the highlight target and the image background are effectively and reasonably displayed in the display equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of infrared image processing, and in particular relates to an adaptive method for suppressing highlighted targets in infrared images with a wide grayscale range. Background Art

[0002] As infrared imaging systems are increasingly used in military detection and civilian life, people are increasingly demanding infrared image quality and scene adaptability. The original data of infrared images is often 14-bit wide grayscale range data. In the process of infrared image processing, it is usually necessary to convert it into 8-bit image data that is easy to display and transmit through dynamic range compression technology. In the compression process, the contrast and details are improved at the same time, thereby enhancing the display effect.

[0003] Image enhancement, as an important part of image processing, plays an important role in improving image quality. With the continuous deepening of research on image enhancement technology, new image enhancement methods continue to emerge. At present, image enhancement methods are mainly divided into the following categories. Grayscale mapping-based methods are commonly used processing techniques for adjusting the grayscale dynamic range of infrared images. The more commonly used mapping curve types are as follows: linear transformation mapping, logarithmic transformation, gamma correction, and Semi-S curve mapping. Due to the great limitations of grayscale mapping of fixed curves, an enhancement algorithm based on adaptive selection of mapping curves based on different judgment criteria has been developed. This method has low algorithm complexity and high engineering feasibility, but it also has high requirements for the design of decision functions.

[0004] However, in some special scenarios, such as when a high-temperature or hot object enters the scene, the traditional linear enhancement method cannot fully utilize the effective display range of the display device, which will submerge the background information, making the overall image tone too dark, resulting in the background information not being effectively enhanced. Invalid grayscale space will occupy the effective display range. Since the limited display space is not fully and effectively utilized, the display space is wasted, and important image details are not effectively displayed. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] The technical problem to be solved by the present invention is to design a scene-adaptive infrared image highlight suppression method so that both the highlight target and the image background can be effectively and reasonably displayed in a display device.

[0007] (II) Technical solution

[0008] In order to solve the above technical problems, the present invention provides an adaptive method for suppressing highlighted targets in infrared images with a wide grayscale range, comprising the following steps:

[0009] Step 1: Get the original infrared image;

[0010] Step 2: Perform non-uniform correction and bad pixel replacement on the original infrared image;

[0011] Step 3: Perform histogram statistics on the processed infrared image, and obtain the image's grayscale value near the upper limit and the grayscale value near the lower limit according to the statistical results. Use the Schmidt trigger threshold to determine whether to enter the adaptive highlight suppression algorithm processing flow. If the start condition is met, go to step 4, if not, go to step 7;

[0012] Step 4: Perform histogram aggregation and compression processing to aggregate the grayscale pixels and make room for image display for other grayscale areas;

[0013] In the design of the compression histogram, two thresholds need to be controlled. One is the grayscale control threshold T ju d ge , if the gray value statistics are less than this threshold, the gray value statistics of the next level are aggregated to the current gray level, and the subsequent gray value statistics are advanced forward; if the cumulative statistics of the gray values ​​of the current gray level and the next gray level still do not exceed the cumulative threshold T sum , then continue to aggregate the grayscale value statistics of the next level to the current grayscale level until it exceeds the cumulative threshold T sum ; If the cumulative value of the previous merged grayscale spaces is greater than T sum , then consider the next step of merging in the next gray space; and so on, regroup and compress the original histogram distribution to obtain the compressed histogram statistics hist′(j);

[0014] Step 5: Perform histogram statistics on the processed infrared image, and obtain the following information based on the histogram statistics: the infrared image effective dynamic range histRange′, the maximum effective gray level histMax′, and the minimum effective gray level histMin′;

[0015] Step 6: Based on step 5, a curve mapping method is used to replace the linear straight line mapping, so as to perform nonlinear mapping processing on the infrared image;

[0016] Step 7: Perform linear mapping on the infrared image.

[0017] The present invention also provides a system for implementing the method.

[0018] (III) Beneficial effects

[0019] The present invention utilizes the histogram information of the original image, and adaptively determines whether to enter the highlight suppression algorithm processing flow according to the statistical results and the decision function. In the process, the histogram of the image is aggregated and compressed, and a nonlinear mapping method is used at the same time to achieve the purpose of effectively suppressing the highlight target, so that the highlight target and the image background are effectively and reasonably displayed in the display device. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a flow chart of the method of the present invention;

[0021] Figure 2 Flowchart for histogram aggregation and compression;

[0022] Figure 3 Comparison of processing effects when a highlighted target enters the scene. DETAILED DESCRIPTION

[0023] In order to make the purpose, content and advantages of the present invention more clear, the specific implementation methods of the present invention are further described in detail below in conjunction with the drawings and examples.

[0024] The present invention provides a scene-adaptive infrared image highlight suppression method, comprising the following steps:

[0025] Step 1: Get the original infrared image;

[0026] Step 2: Perform non-uniform correction and bad pixel replacement on the original infrared image to filter out the influence of bad pixels on histogram statistics;

[0027] Step 3: For simple scenes with a small grayscale range, the noise becomes very obvious due to the large display gain. Therefore, a dynamic range threshold judgment is added on this basis to alleviate the noise suppression problem of simple scenes to a certain extent. For simple scenes with a small grayscale range, there are generally not many unused grayscale levels. If a wide grayscale image display algorithm is used for calculation, it is equivalent to erasing more details, compressing the grayscale range, and further increasing the display gain, making the originally inconspicuous noise become particularly obvious. Therefore, it is very necessary for this algorithm to adaptively judge whether to enter the adaptive highlight suppression algorithm processing flow based on a reasonable decision function. The specific method is: perform histogram statistics on the processed infrared image, and obtain the image's adjacent upper limit grayscale value and adjacent lower line grayscale value information based on the statistical results. Use the Schmidt trigger threshold to determine whether to enter the adaptive highlight suppression algorithm processing flow. If the start condition is met, enter step four, if not, enter step seven;

[0028] Step 4: Perform histogram aggregation and compression processing to aggregate the grayscale pixels, thereby achieving the purpose of compressing the image histogram and making room for image display for other grayscale areas.

[0029] In the design of the compression histogram, two thresholds need to be controlled. One is the grayscale control threshold T ju d ge , if the gray value statistics are less than this threshold, the gray value statistics of the next level are aggregated to the current gray level, and the subsequent gray value statistics are advanced forward; if the cumulative statistics of the gray values ​​of the current gray level and the next gray level still do not exceed the cumulative threshold T sum , then continue to aggregate the grayscale value statistics of the next level to the current grayscale level until it exceeds the cumulative threshold T sum ; If the cumulative value of the previous merged grayscale spaces is greater than T sum , then consider the next step of merging in the next gray space; and so on, re-aggregate and compress the original histogram distribution to obtain the compressed histogram statistics hist′(j); T judge Generally, the value is relatively small, and it is recommended to take a value below 10. sum Take the average number of pixels with each gray value in an image.

[0030] Step 5: Perform histogram statistics on the processed infrared image, and obtain the following information based on the histogram statistics results: including the effective dynamic range of the infrared image histRange′ (the effective dynamic range of the infrared image after histogram collection processing), the maximum effective grayscale level histMax′ (the maximum effective grayscale value of the image after histogram collection processing), and the minimum effective grayscale level histMin′ (the minimum effective grayscale value of the image after histogram collection processing).

[0031] Step 6: Perform nonlinear mapping on the infrared image. The main idea is to use the curve mapping method to replace the linear straight line mapping (the first straight line mapping method), so as to effectively control the suppression effect of the strong target on the image background.

[0032] First, according to the infrared image effective dynamic range histRange′ and the grayscale range Y of the target mapping threshold range Calculate the compression coefficient R, and then get the bidirectional mapping curve according to formula (5) and formula (6). According to the average grayscale avg of the image, count the number of pixels less than the average grayscale. Use formula (7) to divide the statistical number by the image size to get the pixel ratio λ of the low grayscale area. Finally, use the λ parameter as the weighting coefficient and get the final mapping curve according to formula (8). Through this mapping curve, the processing result of adaptive nonlinear mapping can be obtained to achieve the purpose of highlight suppression.

[0033] Step 7: If it is determined that the dynamic range does not exceed the opening threshold, the infrared image is linearly mapped.

[0034] In step 3, the method for determining whether to enter the processing flow of the adaptive highlight suppression algorithm is as follows: according to the histogram statistical results, the grayscale value near the lower limit histDown and the grayscale value near the upper limit histUp are calculated respectively. The histDown needs to start from the lowest grayscale value and accumulate the grayscale quantity from low to high. When the accumulated value is greater than 0.5% of the number of pixels in the whole image, the current grayscale value is recorded as the grayscale value near the lower limit; the histUp needs to start from the highest value and accumulate the grayscale quantity from high to low. When the accumulated value is greater than 0.5% of the number of pixels in the whole image, the current grayscale value is recorded as the grayscale value near the upper limit;

[0035] According to the formula, the algorithm selection is judged, and S(n) represents the judgment result of the current low n frames, where H close To turn off the adaptive nonlinear algorithm threshold, H open To start the adaptive nonlinear algorithm threshold, H close <H open ; When the difference between the gray value near the upper limit and the gray value near the lower limit is less than H close When S(n) = 0, it means that the linear mapping algorithm is performed; when the difference between the grayscale value near the upper limit and the grayscale value near the lower limit is greater than H open When S(n) = 1, it means that the nonlinear mapping algorithm is performed. When the gray value near the upper limit and the gray value near the lower limit are between H close With H open When the selection state of the previous frame is maintained, as shown in the following formula (1), this type of Schmitt triggering method avoids the phenomenon of image flickering caused by the image fluctuating just around a single threshold.

[0036]

[0037] In step 4, calculate T sum The formula is

[0038]

[0039] Where: W is the image width (in pixels);

[0040] H is the image height (in pixels);

[0041] X range is the grayscale range of the original image (unitless).

[0042] 1) Starting from the minimum gray level, determine whether the original gray level statistical pixel value is less than the gray level control threshold T judge ;

[0043] 2) If the original grayscale statistical pixel value is greater than the grayscale control threshold, the new grayscale is increased by one, and the number of pixels is the number of grayscales; if the original grayscale statistical pixel value is less than the grayscale control threshold Tjudge , starting from this gray level, the cumulative sum is calculated step by step. When the cumulative sum exceeds T sum When , the new gray level increases by 1, and the number of pixels is the cumulative sum of the number of pixels;

[0044] 3) Compare the grayscale from small to large step by step according to the principles of 1) and 2), and finally obtain the new histogram statistical result after histogram aggregation and compression.

[0045] In step 6, the formula for calculating the compression coefficient R is as follows:

[0046]

[0047] Y range The grayscale range of the target mapping threshold. If it is an 8-bit display, Y range That is 256. If it is a 10-bit display, Y range That is 1024; the compression factor R can be adaptively adjusted according to the range of the processed image.

[0048] Since the curve model Y1 cannot compress the distribution of low grayscale areas, we map the model in the opposite direction to obtain another mapping curve model Y2. The curve model Y1 is used to compress high grayscale areas and stretch low grayscale areas; the mapping curve model Y2 is used to compress low grayscale areas and stretch high grayscale areas. Therefore, according to the pixel number ratio of high grayscale areas and low grayscale areas decomposed above, the weighted average of Y1 and Y2 can be used to obtain a new mapping curve Y, which becomes our final implementation form. The specific implementation form of the model is as follows:

[0049]

[0050]

[0051] In the formula, Y1 is the forward mapping result, Y2 is the reverse mapping result, and x is the grayscale of the original image. After obtaining the two mapping curve models, the proportion of the low grayscale area segmented by the grayscale mean to the entire image is calculated as λ; according to the average grayscale avg of the image, the number of pixels less than the average grayscale is counted, and the number of pixels counted is divided by the image size to get λ. The calculation formula is as follows:

[0052]

[0053] Where: hist′(j) is the new histogram distribution after the histogram aggregation and compression processing, expressed as the statistical number of gray level j, Represents the sum of the statistical quantities of grayscale below the mean;

[0054] Finally, the parameter λ is used as the weighting coefficient, and the final mapping curve is obtained according to formula (8). Through this mapping curve, the processing result of adaptive nonlinear mapping can be obtained to achieve the purpose of highlight suppression.

[0055] Y=λY1+(1-λ)(Y range -Y2)……………………(8)

[0056] In step 7, if it is determined that the dynamic range does not exceed the opening threshold, linear mapping is performed according to equations (9) and (10), and overflow protection is performed on the calculation result.

[0057] Y=gain(x-avg)+Y range / 2……………………(9)

[0058]

[0059] In order to verify the applicability of the algorithm to the image display of high-temperature targets, we compared the processing effects of the algorithm to a certain extent for some typical scenes. At the same time, we compared the images before and after optimization, and compared the high-temperature objects entering the scene with the low-temperature objects entering the scene. At the same time, we gave the histogram of the image changes and the comparison of the mapping curves before and after optimization. In this way, the feasibility of the algorithm and the comparison before and after the algorithm optimization were verified.

[0060] The method of the present invention uses the histogram information of the original image to reasonably adjust the histogram of the image, and determines whether to start the highlight suppression algorithm through the histogram information, so as to achieve the purpose of adaptively suppressing highlight targets within a wide temperature range. Whether to enable the wide grayscale image display algorithm is determined based on the difference between the grayscale value near the lower limit histDown and the grayscale value near the upper limit histUp. A trigger design similar to the Schmidt trigger is set, and the phenomenon of image flickering caused by algorithm switching near a single threshold is also avoided.

[0061] The adaptive nonlinear mapping curve model of the algorithm is an inverse S-curve, which can better utilize the limited display space of the display device and fully display the suppressed image details. The correctness and applicability of the algorithm model are verified by comparing the processing of typical scene images, so that the highlight target and image background can be effectively and layeredly displayed in the display device in different scenes.

[0062] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An adaptive method for suppressing highlighted targets in infrared images with a wide grayscale range, characterized in that: The following steps are involved: Step 1: Get the original infrared image; Step 2: Perform non-uniform correction and bad pixel replacement on the original infrared image; Step 3: Perform histogram statistics on the processed infrared image, and obtain the image's grayscale value near the upper limit and the grayscale value near the lower limit according to the statistical results. Use the Schmidt trigger threshold to determine whether to enter the adaptive highlight suppression algorithm processing flow. If the start condition is met, go to step 4, if not, go to step 7; Step 4: Perform histogram aggregation and compression processing to aggregate the grayscale pixels and make room for image display for other grayscale areas; In the design of the compression histogram, two thresholds need to be controlled. One is the grayscale control threshold T ju d ge , if the gray value statistics are less than this threshold, the gray value statistics of the next level are aggregated to the current gray level, and the subsequent gray value statistics are advanced forward; if the cumulative statistics of the gray values ​​of the current gray level and the next gray level still do not exceed the cumulative threshold T sum , then continue to aggregate the grayscale value statistics of the next level to the current grayscale level until it exceeds the cumulative threshold T sum ; If the cumulative value of the previous merged grayscale spaces is greater than T sum , then consider the next step of merging in the next gray space; and so on, regroup and compress the original histogram distribution to obtain the compressed histogram statistics hist′(j); Step 5: Perform histogram statistics on the processed infrared image, and obtain the following information based on the histogram statistics: the infrared image effective dynamic range histRange′, the maximum effective gray level histMax′, and the minimum effective gray level histMin′; Step 6: Based on step 5, a curve mapping method is used to replace the linear straight line mapping, so as to perform nonlinear mapping processing on the infrared image; Step 7: Perform linear mapping on the infrared image.

2. The method according to claim 1, characterized in that In step 3, the method for determining whether to enter the processing flow of the adaptive highlight suppression algorithm is as follows: according to the histogram statistical results, the grayscale value near the lower limit histDown and the grayscale value near the upper limit histUp are calculated respectively. The histDown needs to start from the lowest grayscale value and accumulate the grayscale quantity from low to high. When the accumulated value is greater than 0.5% of the number of pixels in the whole image, the current grayscale value is recorded as the grayscale value near the lower limit; the histUp needs to start from the highest value and accumulate the grayscale quantity from high to low. When the accumulated value is greater than 0.5% of the number of pixels in the whole image, the current grayscale value is recorded as the grayscale value near the upper limit; Let S(n) represent the decision result of the current lower n frames, where H close To turn off the adaptive nonlinear algorithm threshold, H open To start the adaptive nonlinear algorithm threshold, H close <H open ; When the difference between the gray value near the upper limit and the gray value near the lower limit is less than H close When S(n) = 0, it means that the linear mapping algorithm is performed; when the difference between the grayscale value near the upper limit and the grayscale value near the lower limit is greater than H open When S(n) = 1, it means that the nonlinear mapping algorithm is performed. When the gray value near the upper limit and the gray value near the lower limit are between H close With H open When the selection state of the previous frame is maintained, as shown in the following formula (1):

3. The method according to claim 2, characterized in that In step 4, calculate T sum The formula is Where: W is the image width; H is the image height; X range is the grayscale range of the original image; 1) Starting from the minimum gray level, determine whether the original gray level statistical pixel value is less than the gray level control threshold T judge ; 2) If the original grayscale statistical pixel value is greater than the grayscale control threshold, the new grayscale is increased by one, and the number of pixels is the number of grayscale levels; If the original grayscale statistical pixel value is less than the grayscale control threshold T judge , starting from this gray level, the cumulative sum is calculated step by step. When the cumulative sum exceeds T sum When , the new gray level increases by 1, and the number of pixels is the cumulative sum of the number of pixels; 3) Compare the grayscale from small to large step by step according to the principles of 1) and 2), and finally obtain the new histogram statistical result after histogram aggregation and compression.

4. The method according to claim 3, characterized in that In step 6, firstly, according to the infrared image effective dynamic range histRange′ and the grayscale range Y of the target mapping threshold range Calculate the compression factor R; Then the bidirectional mapping curve is obtained, and according to the average grayscale avg of the image, the number of pixels less than the average grayscale is counted, and the statistical number is divided by the image size to obtain the pixel ratio λ of the low grayscale area. Finally, the λ parameter is used as the weighting coefficient to obtain the final mapping curve. Through this mapping curve, the processing result of adaptive nonlinear mapping is obtained.

5. The method according to claim 4, characterized in that In step 6, the formula for calculating the compression coefficient R is as follows: Y range The grayscale range for the target mapping threshold.

6. The method according to claim 5, characterized in that In step 6, the bidirectional mapping curve is obtained according to equations (5) and (6): Where Y1 is the forward mapping result, Y2 is the reverse mapping result, and x is the grayscale of the original image.

7. The method according to claim 6, characterized in that In step 6, according to the average grayscale avg of the image, the number of pixels with a grayscale lower than the average grayscale is counted, and the pixel ratio λ of the low grayscale area is obtained by dividing the counted number by the image size using formula (7): Where: hist′(j) is the new histogram distribution after the histogram aggregation and compression processing, expressed as the statistical number of gray level j, Represents the sum of the statistical quantities of grayscale below the mean.

8. The method according to claim 7, characterized in that In step 6, the parameter λ is used as the weighting coefficient, and the final mapping curve is obtained according to formula (8). Through this mapping curve, the processing result of the adaptive nonlinear mapping is obtained: Y=λY1+(1-λ)(Y range -Y2)……………………(8)。 9. The method according to claim 8, characterized in that In step 7, linear mapping is performed according to equations (9) and (10), and overflow protection is performed on the calculation results. Y=gain(x-avg)+Y range / 2……………………(9) 10. A system for implementing the method according to any one of claims 1 to 9.

Images