A deep learning side face image processing method and system based on light compensation

Abstract

The invention discloses a light compensation-based deep learning side face image processing method and system, belonging to the technical field of image processing, including: S1, constructing a data set; S101, cutting out an original image to obtain a side face area and a background image region; S102, mark the side face region as 1; S103, according to the size of the side face region, in the background image region, randomly select a comparison image of the same size, and mark the comparison image as 0; S104, repeat N times S103, obtain N comparison images; S105, replace the original image, repeat S101-S104, and obtain the side face recognition data set; S2, perform light compensation on the cutout image in the side face recognition data set through the artificial neural network, and obtain the compensated image; S3 . Profile face recognition; S301. Divide the compensated image into M regional sub-images; S302. Perform profile face recognition on each regional sub-image to obtain a profile face image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a light compensation-based deep learning side face image processing method and system. Background technique [0002] As we all know, face recognition is a biometric technology for identification based on human facial feature information. A series of related technologies that use a video camera or camera to collect images or video streams containing human faces, automatically detect and track human faces in the images, and then perform facial recognition on the detected faces, usually also called portrait recognition and facial recognition. . [0003] At present, the face recognition system mainly includes image capture, face positioning, image preprocessing, and face recognition (identity confirmation or identity search). Among them: the quality of image capture will directly affect the later stage: the difficulty of image processing and the accuracy of the...

AI Technical Summary

Problems solved by technology

[0004] In the actual image collection process, sometimes it is difficult to obtain a complete and clear image. For example, some criminals often avoid the camera when committing crimes. Only the side information of the person is obtained; and it is often in a dark corner. For these low-quality images, in order to identify the side face that is blocked in the dark corner, the traditional technology is to identify the side face area in the entire image, and the side face Deviation correction is based on image interpolation, which usually uses complete comparison images of frontal and side faces from different angles as a data set. These methods do not consider the background and application scenarios. It is extremely difficult for the scene where the target person only shows part of the side face in the dark corner correct identification

The experiment found that it is difficult for this traditional face recognition technology to obtain recognition results quickly and accurately;

Images