A Video Face Recognition Method Based on Incremental Learning of Face Segmented Visual Representation

Abstract

The invention provides a video face recognition method based on incremental learning of face segmentation visual representation, which belongs to the field of pattern recognition. This method uses the Adaboost algorithm to detect the frontal face image of the first frame of the face video, uses the Camshift algorithm to track, and obtains all face images. In the process of reading the video, the face images are incrementally clustered, and Select representatives from each type of face images; process the representative images to learn a visual dictionary based on block visual representation; use the visual dictionary to characterize the face images; finally recognize the video composed of face images according to the similarity matrix. The method of the invention can improve the recognition rate and robustness of the human face in the video under the conditions of unsatisfactory illumination, posture and tracking result, and can effectively, conveniently and automatically detect, track and recognize the human face in the video.

Description

technical field [0001] The invention belongs to the field of pattern recognition, and relates to technologies such as image processing and computer vision, in particular to a face recognition method based on incremental learning of face block visual representation. Background technique [0002] Face recognition in video is mainly for the analysis and processing of moving image sequences containing people. The face recognition problem can be defined as: input (query) still images or videos in the scene, and use the face database to identify or verify one of the scenes. one or more people. Face recognition based on still images usually refers to inputting (querying) a still image and using a face database to identify or verify the face in the image. Video-based face recognition refers to inputting (querying) a video and using a face database to identify or verify the face in the video. [0003] Category is a basic attribute of all things in the world. Things of the same cate...

AI Technical Summary

Problems solved by technology

At present, many algorithms have been studied on face recognition (classification) in video scenes, but they often have certain shortcomings. For example, the collection and labeling of the database is required, the training samples need to be retrained, and incremental updates cannot be performed.

In addition, due to the large degree of change in the face posture in the video and the influence of external factors such as lighting, some recognition (classification) algorithms can achieve good performance under certain conditions, but they often fail under complex environmental conditions. Unable to achieve good recognition results, detection and tracking results are not ideal

Images