CN-121600012-B - Object identification method, system, equipment and medium based on partition gain

CN121600012BCN 121600012 BCN121600012 BCN 121600012BCN-121600012-B

Abstract

The application discloses an object identification method, system, equipment and medium based on partition gain, belonging to the technical field of target identification; the method comprises the steps of acquiring image information, carrying out target edge detection on the acquired image information to obtain image information containing edge information, dividing an original image into a plurality of non-overlapping area blocks according to the edge information in the image, carrying out image gain on the area blocks respectively, recombining the area blocks after gain into complete image information, and outputting the image information after gain. In addition, the method also avoids edge missing detection or false detection caused by the next detection of low illumination by judging the overlapping degree of edge information for a plurality of times and performing iterative detection, thereby enhancing the robustness of the processed image.

Inventors

Request for anonymity

Assignees

创视半导体(杭州)有限公司

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (9)

1. An object recognition method based on zoning gain, comprising: collecting image information of a region to be detected in real time; Performing target edge detection on the acquired image information to obtain image information containing edge information; Dividing an original image into a plurality of non-overlapping area blocks according to edge information in the image, respectively carrying out image gain on the plurality of area blocks, and recombining the plurality of area blocks after gain into complete image information; outputting the image information after gain; the method for dividing the original image into a plurality of non-overlapping area blocks and respectively carrying out image gain on the plurality of area blocks comprises the following steps: marking an original image according to each target image according to the edge information, and dividing the original image into a plurality of non-overlapping area blocks; Dividing a corresponding region in a preset image into a plurality of region blocks corresponding to the region blocks according to the plurality of region blocks of the original image; Each original image area block is subjected to partition gain processing respectively, and the method comprises the following steps: Calculating the average brightness and contrast of the original image area block, and calculating the average brightness and contrast of the corresponding preset image area block; Respectively obtaining image evaluation parameters of an original image area block and image evaluation parameters of corresponding preset image area blocks according to preset average brightness weights and contrast weights in different scenes and obtaining image evaluation parameter ratios; and comparing the image evaluation parameter ratio with an expected value thereof, and performing gain processing on the original image area block according to a comparison result.
2. The method for recognizing an object based on partition gain according to claim 1, wherein said comparing the image evaluation parameter with its expected value, gain-processing the original image area block according to the comparison result, comprises: If the image evaluation parameter ratio is smaller than or equal to the expected value, performing brightness gain on the original image area block, otherwise, directly outputting the original image area block information; When the image evaluation parameter ratio is smaller than or equal to an expected value, calculating the ratio of the image average brightness of the original image area block to the image average brightness of the corresponding preset image area block, and calculating the ratio of the image contrast of the original image area block to the image contrast of the corresponding preset image area block; According to the ratio of the average brightness of the image to the contrast of the image, and the preset average brightness adjustment coefficient of the image and the preset contrast adjustment coefficient of the image, adjusting to obtain the average brightness after the block gain and the contrast after the gain of the original image area; Calculating an image evaluation parameter after the gain of the original image area block according to the average brightness after the gain and the contrast after the gain of the original image area block, calculating the image evaluation parameter ratio after the gain according to the image evaluation parameter ratio after the gain, comparing the image evaluation parameter ratio after the gain with an expected value of the image evaluation parameter ratio after the gain, if the image evaluation parameter ratio after the gain is still smaller than or equal to the expected value of the image evaluation parameter ratio after the gain, returning to a previous step to iteratively adjust the average brightness after the gain and the contrast after the gain of the original image area block until the image evaluation parameter ratio after the gain is larger than the expected value of the image evaluation parameter ratio after the gain, and performing inverse operation of average brightness and saturation according to the average brightness after the gain of the original image area block at the moment to obtain brightness gain and contrast gain, and performing gain processing on the original image area block.
3. The method for identifying an object based on zoned gain according to any one of claims 1-2, further comprising, prior to said outputting the gain-after-image information: The image information after the partition gain is subjected to region detection judgment, and the region detection judgment process comprises the following steps: according to the obtained target area in the image information, a plurality of units are extended to the periphery by taking the target area as the center, and target edge detection is carried out again according to the extended image; comparing the edge information obtained by current detection with the edge information obtained by previous detection, and calculating to obtain the coincidence ratio of the current edge information and the edge information obtained by previous detection; comparing the overlap ratio with a preset threshold value, if the overlap ratio is larger than or equal to the preset threshold value, judging that the image edge identification processing is correct at the moment, obtaining and outputting an image containing one or more edge information, otherwise, returning to the extension processing step for carrying out next target edge detection.
4. A method for identifying objects based on zoned gain according to claim 3, wherein said outputting the gain-based image information comprises: If the acquired single frame image is the image, outputting the image with correct edge identification processing as image information with a target object detection frame; If the video information is collected, the image frames in the video information are processed according to time sequence, and the image with correct edge recognition processing is output as the image information with the target object detection frame.
5. The zoned gain-based object recognition method of claim 4, further comprising, if video information is collected: extracting target object feature points in the image according to the correct image processed by edge recognition; And acquiring coordinates of the same feature points in different image frames in an image coordinate system, and judging whether the target object moves or not according to the image coordinate information of the same feature points in different image frames.
6. An object recognition system based on zoned gain, comprising: The image acquisition module is used for acquiring image information of the region to be detected in real time and transmitting the image information to the object identification module; The object identification module is used for carrying out target edge detection on the collected image information, obtaining the image information containing the edge information and transmitting the image information to the partition gain module; The partition gain module is used for dividing an original image into a plurality of non-overlapping area blocks according to edge information in the image, respectively carrying out image gain on the plurality of area blocks, recombining the plurality of area blocks after gain into complete image information and transmitting the complete image information to the image output module; the image output module is used for outputting the image information after gain; the method for dividing the original image into a plurality of non-overlapping area blocks and respectively carrying out image gain on the plurality of area blocks comprises the following steps: marking an original image according to each target image according to the edge information, and dividing the original image into a plurality of non-overlapping area blocks; Dividing a corresponding region in a preset image into a plurality of region blocks corresponding to the region blocks according to the plurality of region blocks of the original image; Each original image area block is subjected to partition gain processing respectively, and the method comprises the following steps: Calculating the average brightness and contrast of the original image area block, and calculating the average brightness and contrast of the corresponding preset image area block; Respectively obtaining image evaluation parameters of an original image area block and image evaluation parameters of corresponding preset image area blocks according to preset average brightness weights and contrast weights in different scenes and obtaining image evaluation parameter ratios; and comparing the image evaluation parameter ratio with an expected value thereof, and performing gain processing on the original image area block according to a comparison result.
7. The zoned gain-based object recognition system of claim 6, wherein the image output module further comprises: And the region detection judging unit is used for carrying out region detection judgment on the image information after the partition gain before outputting the image, outputting the image passing the region detection judgment, and inputting the image which does not pass the region detection judgment to the object recognition module for carrying out target edge detection again.
8. An electronic device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the object recognition method of any one of claims 1-5 when executing the computer program.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the object recognition method according to any one of claims 1-5.

Description

Object identification method, system, equipment and medium based on partition gain Technical Field The application belongs to the technical field of target recognition, and particularly relates to an object recognition method, system, equipment and medium based on partition gain. Background With the development of intelligence, the requirements of the image sensor field for detecting moving objects are increasing. Although the prior art can accurately identify the moving object, the method still has the problem of lower accuracy for identifying and detecting the moving object with low illumination. In the prior art, a method for improving the recognition accuracy of a moving object under low illumination by using color brightness to gain the object under low illumination is proposed, but after image gain is performed, the technology may cause problems of inconspicuous image gain or erroneous recognition caused by color distortion, and in addition, the technology still has the conditions of lower accuracy and incapability of being universally adapted to object detection and recognition under the low illumination environment. Disclosure of Invention Aiming at the problems in the prior art, the application provides an object identification method, system, equipment and medium based on partition gain, which are used for adaptively adjusting the gain of a target area by detecting the edge of an object, and carrying out object detection based on the target area after the gain, thereby improving the accuracy of object identification in a low-illumination environment. The application is realized by the following technical scheme: An object identification method based on zoning gain, comprising: collecting image information of a region to be detected in real time; Performing target edge detection on the acquired image information to obtain image information containing edge information; Dividing an original image into a plurality of non-overlapping area blocks according to edge information in the image, respectively carrying out image gain on the plurality of area blocks, and recombining the plurality of area blocks after gain into complete image information; and outputting the image information after gain. In some embodiments, the dividing the original image into a plurality of non-overlapping region blocks and performing image gain on the plurality of region blocks respectively includes: marking an original image according to each target image according to the edge information, and dividing the original image into a plurality of non-overlapping area blocks; Dividing a corresponding region in a preset image into a plurality of region blocks corresponding to the region blocks according to the plurality of region blocks of the original image; Each original image area block is subjected to partition gain processing respectively, and the method comprises the following steps: Calculating the average brightness and contrast of the original image area block, and calculating the average brightness and contrast of the corresponding preset image area block; Respectively obtaining image evaluation parameters of an original image area block and image evaluation parameters of corresponding preset image area blocks according to preset average brightness weights and contrast weights in different scenes and obtaining image evaluation parameter ratios; and comparing the image evaluation parameter ratio with an expected value thereof, and performing gain processing on the original image area block according to a comparison result. In some embodiments, comparing the image evaluation parameter with its expected value, performing gain processing on the original image area block according to the comparison result, including: If the image evaluation parameter ratio is smaller than or equal to the expected value, performing brightness gain on the original image area block, otherwise, directly outputting the original image area block information; When the image evaluation parameter ratio is smaller than or equal to an expected value, calculating the ratio of the image average brightness of the original image area block to the image average brightness of the corresponding preset image area block, and calculating the ratio of the image contrast of the original image area block to the image contrast of the corresponding preset image area block; According to the ratio of the average brightness of the image to the contrast of the image, and the preset average brightness adjustment coefficient of the image and the preset contrast adjustment coefficient of the image, adjusting to obtain the average brightness after the block gain and the contrast after the gain of the original image area; Calculating an image evaluation parameter after the gain of the original image area block according to the average brightness after the gain and the contrast after the gain of the original image area block, calculating the image evaluation parameter ratio after t