An adaptive dual-recording video exposure assessment method based on histogram fusion

Abstract

The present invention discloses an adaptive dual-recording video picture exposure assessment method based on histogram fusion, which relates to the technical field of dual-recording video quality detection. By dividing the local key areas of the video picture, the global brightness image and the local brightness image are uniformly scaled to obtain a normalized global brightness image and a normalized local brightness image, which is conducive to performing histogram probability distribution statistics on the same pixel scale, avoiding the noise disturbance of pixel data sample values ​​and the influence of the sampling number on the histogram. The weighted superposition of the global and local brightness histograms is combined, and a probability truncation operation is introduced to eliminate noise interference, and finally a comprehensive exposure quality score is calculated by fusing the overexposure and underexposure deviations. The present invention solves the limitations of relying solely on global or local brightness analysis, significantly improves the exposure quality inspection accuracy of key areas in dual-recording videos, and reduces the influence of noise on statistical distribution through truncation correction.

Description

Technical Field

[0001] The present invention relates to the technical field of dual-recording video quality detection, and in particular to an adaptive dual-recording video picture exposure assessment method based on histogram fusion. Background Art

[0002] With advancements in digital video software and hardware technology and significant cost reductions, digital video applications are being adopted by a growing number of industries and fields. Financial institutions such as banks, securities firms, and insurance companies are required by regulatory authorities to conduct both audio and video recording (referred to as dual recording) when selling wealth management products, precious metals, and insurance products. Furthermore, dual recording must comply with a series of regulatory standards and requirements. As a digital medium for display, dual recording must ensure that both the audio and video streams are properly displayed and played. This proper playback typically requires clear audio and video content, smooth playback without interruptions, and synchronization. The clarity of audio and video content is a key indicator of video quality, which is closely related to video exposure quality. Exposure quality directly impacts the visual quality and readability of the video. In dual recording, the exposure quality of key areas (such as faces) is particularly crucial.

[0003] However, traditional video exposure quality inspection methods often suffer from the following problems: First, they typically rely solely on global or local brightness analysis, failing to effectively consider both global and local brightness characteristics, resulting in low exposure quality inspection accuracy in critical areas. Second, traditional statistical analysis methods are susceptible to noise, affecting the accuracy of exposure quality scores. To address these issues, manual inspection can address some of these needs, but with large video volumes, it undoubtedly puts a severe strain on human energy and endurance. Furthermore, manual review inevitably suffers from a series of drawbacks, including low efficiency, poor accuracy, inconsistent judgment criteria, and high costs. To address this issue, automated, intelligent exposure quality inspection of dual-recorded videos using computers has become imperative. Summary of the Invention

[0004] In order to solve the above-mentioned technical problems of video exposure quality inspection, the present invention provides an adaptive dual-recording video exposure assessment method based on histogram fusion. The following technical solutions are adopted:

[0005] A method for adaptive dual-recording video exposure assessment based on histogram fusion, comprising the following steps:

[0006] Step 1: Convert the video image to be tested into a brightness image and define the key area for exposure detection;

[0007] Step 2: uniformly scaling the global brightness image and the local brightness image to obtain a normalized global brightness image and a normalized local brightness image;

[0008] Step 3: Calculate a global histogram from the global normalized brightness image, calculate a local histogram from the local normalized brightness image, perform weighted superposition of the global histogram and the local histogram to obtain a superimposed histogram, and then truncate the superimposed histogram to obtain a truncated histogram;

[0009] Step 4, calculating the exposure separation of the histogram obtained in step 3, dividing the exposure deviation into overexposure deviation and underexposure deviation, and calculating the overexposure deviation and underexposure deviation respectively;

[0010] Step 5: Calculate the exposure of the histogram based on the overexposure deviation and the underexposure deviation. The exposure value is between 0 and 1. The exposure of the video image to be detected is evaluated based on the position of the exposure between 0 and 1.

[0011] By employing this technical solution, the video image is divided into key local areas (such as the face area) and the global and local luminance images are uniformly scaled to s×s dimensions. The resulting normalized global and local luminance images facilitate histogram probability distribution statistics at the same pixel scale, avoiding the effects of pixel data sample noise and the number of samples on the histogram.

[0012] This method combines a weighted superposition of global and local brightness histograms and introduces a probabilistic truncation operation to eliminate noise interference. Ultimately, it calculates a comprehensive exposure quality score by integrating overexposure and underexposure deviations. Compared to traditional methods, this method overcomes the limitations of relying solely on global or local brightness analysis, significantly improving the exposure quality inspection accuracy of key areas (such as faces) in dual-recorded videos. Truncation correction also reduces the impact of noise on the statistical distribution.

[0013] Optionally, in step 3, the formula for calculating the global histogram is:

[0014] ;

[0015] in Get the pixel value. ; is the pixel value in the global normalized brightness image The probability of occurrence;

[0016] The formula for calculating the local histogram is:

[0017] ;

[0018] in Get the pixel value. ; is the pixel value in the local normalized brightness image probability of occurrence.

[0019] Optionally, in step 3, the calculation formula for the overlay histogram is:

[0020]

[0021] in Get the pixel value. ; for and The superposition value of The importance weight of the local area, ranging from 0 to 1, is generally .

[0022] By employing this technical solution, local region importance weighting can effectively adjust the exposure focus area, overcoming the shortcomings of traditional exposure assessment algorithms that rely solely on global or local brightness. This is particularly true in areas such as dual-recording video, where the exposure quality of key areas such as faces is comprehensively assessed alongside the exposure quality of the entire frame.

[0023] Optionally, the formula for truncating the overlay histogram is as follows:

[0024] ;

[0025] in Represents a truncation operation, which takes all distribution probabilities less than The probability value of is set to 0 and is defined as follows:

[0026]

[0027] By adopting the above technical solution, the truncation operation can effectively overcome the disturbance of noise pixels on the brightness histogram distribution, which is beneficial to improving the accuracy of the algorithm.

[0028] Optionally, the overexposure deviation is a ratio of a minimum pixel value in the histogram to 256; and the underexposure deviation is a ratio of a value obtained by subtracting a maximum pixel value from 255 in the histogram to 256.

[0029] Optionally, the exposure E is calculated as follows:

[0030] ;

[0031] Where E is the exposure, is the overexposure deviation, is the underexposure deviation.

[0032] Optionally, E<0.35 indicates that the image is underexposed, E>0.65 indicates that the image is overexposed, and in other cases the exposure is normal.

[0033] By adopting the above technical solution, the two values ​​of underexposure and overexposure can be merged into one variable, and the value range is between 0 and 1. The closer the value is to 0, the more underexposed the picture is, and the closer it is to 1, the closer the picture is to overexposed.

[0034] A histogram fusion-based adaptive dual-recording video exposure assessment device is used to implement an adaptive dual-recording video exposure assessment method based on histogram fusion. The adaptive dual-recording video exposure assessment device includes a memory and a processor. The memory stores a dual-recording video quality assessment program designed using an adaptive dual-recording video exposure assessment method based on histogram fusion, and stores a video image to be detected. The processor is communicatively connected to the memory, inputs the video image to be detected into the dual-recording video quality assessment program, runs the dual-recording video quality assessment program, and outputs a dual-recording video quality assessment result.

[0035] Optionally, a display is further included, which is communicatively connected to the processor and is used to display the dual-recording video quality assessment result output by the processor.

[0036] The memory stores a dual-recording video quality assessment program designed using an adaptive dual-recording video picture exposure assessment method based on histogram fusion.

[0037] In summary, the present invention includes at least the following beneficial technical effects:

[0038] The present invention can provide an adaptive dual-recording video exposure assessment method based on histogram fusion. By dividing the video image into local key areas (such as the face area), the global brightness image and the local brightness image are uniformly scaled to obtain normalized global brightness image and normalized local brightness image. This is conducive to performing histogram probability distribution statistics on the same pixel scale, avoiding the noise disturbance of pixel data sample values ​​and the influence of the sampling number on the histogram.

[0039] This method combines a weighted superposition of global and local brightness histograms and introduces a probabilistic truncation operation to eliminate noise interference. Ultimately, it calculates a comprehensive exposure quality score by integrating overexposure and underexposure deviations. Compared to traditional methods, this method overcomes the limitations of relying solely on global or local brightness analysis, significantly improving the exposure quality inspection accuracy of key areas (such as faces) in dual-recorded videos. Truncation correction also reduces the impact of noise on the statistical distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flow chart of an adaptive dual-recording video exposure assessment method based on histogram fusion according to the present invention;

[0041] Figure 2 are the global brightness image and the local brightness image in the specific embodiment of the present invention;

[0042] Figure 3 are the normalized global brightness image and the normalized local brightness image in a specific embodiment of the present invention;

[0043] Figure 4 are the global brightness histogram and the local brightness histogram in the specific embodiment of the present invention;

[0044] Figure 5 These are the superimposed brightness histogram and the truncated superimposed brightness histogram in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below with reference to the accompanying drawings. An embodiment of the present invention discloses an adaptive dual-recording video exposure assessment method based on histogram fusion.

[0046] Reference Figure 1-Figure 5 , an adaptive dual-recording video exposure assessment method based on histogram fusion, comprising the following steps:

[0047] Step 1: Convert the video image to be tested into a brightness image and define the key area for exposure detection;

[0048] Step 2: uniformly scaling the global brightness image and the local brightness image to obtain a normalized global brightness image and a normalized local brightness image;

[0049] Step 3: Calculate a global histogram from the global normalized brightness image, calculate a local histogram from the local normalized brightness image, perform weighted superposition of the global histogram and the local histogram to obtain a superimposed histogram, and then truncate the superimposed histogram to obtain a truncated histogram;

[0050] Step 4, calculating the exposure separation of the histogram obtained in step 3, dividing the exposure deviation into overexposure deviation and underexposure deviation, and calculating the overexposure deviation and underexposure deviation respectively;

[0051] Step 5: Calculate the exposure of the histogram based on the overexposure deviation and the underexposure deviation. The exposure value is between 0 and 1. The exposure of the video image to be detected is evaluated based on the position of the exposure between 0 and 1.

[0052] By dividing the video image into key local areas (such as faces), the global and local luminance images are uniformly scaled to s × s dimensions. The resulting normalized global and local luminance images facilitate histogram probability distribution statistics at the same pixel scale, minimizing the effects of pixel data sample noise and the number of samples on the histogram.

[0053] This method combines a weighted superposition of global and local brightness histograms and introduces a probabilistic truncation operation to eliminate noise interference. Ultimately, it calculates a comprehensive exposure quality score by integrating overexposure and underexposure deviations. Compared to traditional methods, this method overcomes the limitations of relying solely on global or local brightness analysis, significantly improving the exposure quality inspection accuracy of key areas (such as faces) in dual-recorded videos. Truncation correction also reduces the impact of noise on the statistical distribution.

[0054] In step 3, the formula for calculating the global histogram is:

[0055] ;

[0056] in Get the pixel value. ; is the pixel value in the global normalized brightness image The probability of occurrence;

[0057] The formula for calculating the local histogram is:

[0058] ;

[0059] in Get the pixel value. ; is the pixel value in the local normalized brightness image probability of occurrence.

[0060] In step 3, the calculation formula for the superimposed histogram is:

[0061]

[0062] in Get the pixel value. ; for and The superposition value of The importance weight of the local area, ranging from 0 to 1, is generally .

[0063] Local area importance weighting effectively adjusts exposure focus areas, overcoming the drawback of traditional exposure assessment algorithms that rely solely on global or local brightness. This is particularly true in areas like dual-recording video, where the exposure quality of key areas like faces is comprehensively assessed alongside the overall exposure quality of the entire frame.

[0064] The formula for truncating the overlay histogram is as follows:

[0065] ;

[0066] in Represents a truncation operation, which takes all distribution probabilities less than The probability value of is set to 0 and is defined as follows:

[0067]

[0068] The truncation operation can effectively overcome the disturbance of noise pixels on the brightness histogram distribution, which is beneficial to improving the accuracy of the algorithm.

[0069] The overexposure deviation is the ratio of the minimum pixel value in the histogram to 256; the underexposure deviation is the ratio of 255 minus the maximum pixel value in the histogram to 256.

[0070] The formula for calculating exposure E is as follows:

[0071] ;

[0072] Where E is the exposure, is the overexposure deviation, is the underexposure deviation.

[0073] E<0.35 indicates that the image is underexposed, E>0.65 indicates that the image is overexposed, and in other cases the exposure is normal.

[0074] The two values ​​of underexposure and overexposure can be combined into one variable, and the value range is between 0 and 1. The closer the value is to 0, the more underexposed the picture is, and the closer it is to 1, the closer the picture is to overexposed.

[0075] A histogram fusion-based adaptive dual-recording video exposure assessment device is used to implement an adaptive dual-recording video exposure assessment method based on histogram fusion. The adaptive dual-recording video exposure assessment device includes a memory and a processor. The memory stores a dual-recording video quality assessment program designed using an adaptive dual-recording video exposure assessment method based on histogram fusion, and stores a video image to be detected. The processor is communicatively connected to the memory, inputs the video image to be detected into the dual-recording video quality assessment program, runs the dual-recording video quality assessment program, and outputs a dual-recording video quality assessment result.

[0076] It also includes a display, which is communicatively connected to the processor and is used to display the dual-recording video quality assessment result output by the processor.

[0077] The memory stores a dual-recording video quality assessment program designed using an adaptive dual-recording video picture exposure assessment method based on histogram fusion.

[0078] The following describes the implementation principle of an adaptive dual-recording video exposure assessment method based on histogram fusion based on an embodiment:

[0079] Step 1: Divide the local key areas of the video image to be detected:

[0080] Convert the image into a brightness image and define the key areas for exposure detection. The global brightness image and local brightness image obtained are as follows: Figure 2 As shown;

[0081] Step 2: Scale the global brightness image and the local brightness image uniformly to 512×512 size, as shown in the following example: Figure 3 As shown;

[0082] Step 3: Calculate the global histogram from the global normalized brightness image , calculate the local histogram from the local normalized brightness image ,like Figure 4 As shown;

[0083] Weighted superposition is used to obtain the superposition histogram, and truncation is performed, and the truncation threshold is , the superimposed brightness histogram and the truncated superimposed brightness histogram are as follows Figure 5 As shown;

[0084] Step 4: The exposure deviation is divided into overexposure deviation and underexposure deviation. The underexposure deviation is the ratio of 255 minus the maximum pixel value in the histogram to 256, that is:

[0085]

[0086] The overexposure deviation is the ratio of the minimum pixel value in the histogram to 256, that is:

[0087]

[0088] Step 5: Exposure Calculation

[0089] Exposure The calculation results are as follows:

[0090]

[0091] The calculation shows that the image exposure assessed by the present invention is about 0.31, which is underexposed. Traditional methods often rely on the global brightness histogram and cannot accurately reflect local exposure anomalies.

[0092] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. An adaptive dual-recording video exposure assessment method based on histogram fusion, characterized in that: The following steps are involved: Step 1: Convert the video image to be tested into a brightness image and define the key area for exposure detection; Step 2: uniformly scaling the global brightness image and the local brightness image to obtain a normalized global brightness image and a normalized local brightness image; Step 3: Calculate a global histogram from the global normalized brightness image, calculate a local histogram from the local normalized brightness image, perform weighted superposition of the global histogram and the local histogram to obtain a superimposed histogram, and then truncate the superimposed histogram to obtain a truncated histogram; Step 4, calculating the exposure separation of the histogram obtained in step 3, dividing the exposure deviation into overexposure deviation and underexposure deviation, and calculating the overexposure deviation and underexposure deviation respectively; Step 5: Calculate the exposure of the histogram based on the overexposure deviation and the underexposure deviation, where the exposure value is between 0 and 1, and evaluate the exposure of the video image to be detected based on the position of the exposure between 0 and 1; In step 3, the formula for calculating the global histogram is: ； in Get the pixel value. ; is the pixel value in the global normalized brightness image The probability of occurrence; The formula for calculating the local histogram is: ； in Get the pixel value. ; is the pixel value in the local normalized brightness image The probability of occurrence; In step 3, the calculation formula for the superimposed histogram is: ； ； in Get the pixel value. ; for and The superposition value of is the importance weight of the local area, ranging from 0 to 1; The formula for truncating the overlay histogram is as follows: ； in Represents a truncation operation, which takes all distribution probabilities less than The probability value of is set to 0 and is defined as follows: ； The formula for calculating exposure E is as follows: ； Where E is the exposure, is the overexposure deviation, is the underexposure deviation.

2. The method for adaptive dual-recording video exposure assessment based on histogram fusion according to claim 1, characterized in that: The overexposure deviation is the ratio of the minimum pixel value in the histogram to 256; the underexposure deviation is the ratio of 255 minus the maximum pixel value in the histogram to 256.

3. The method for adaptive dual-recording video exposure assessment based on histogram fusion according to claim 2, characterized in that: E<0.35 indicates that the image is underexposed, E>0.65 indicates that the image is overexposed, and in other cases the exposure is normal.

4. An adaptive dual-recording video exposure assessment device based on histogram fusion, characterized in that: Used to implement the adaptive dual-recording video picture exposure assessment method based on histogram fusion as described in any one of claims 1-3, the adaptive dual-recording video picture exposure assessment device includes a memory and a processor, the memory stores a dual-recording video quality assessment program designed using the adaptive dual-recording video picture exposure assessment method based on histogram fusion as described in any one of claims 1-3, and stores the video picture to be detected, the processor is communicatively connected to the memory, inputs the video picture to be detected into the dual-recording video quality assessment program, runs the dual-recording video quality assessment program, and outputs the dual-recording video quality assessment result.

5. The device for adaptive dual-recording video exposure assessment based on histogram fusion according to claim 4, characterized in that: It also includes a display, which is communicatively connected to the processor and is used to display the dual-recording video quality assessment result output by the processor.

Images