Court monitoring face recognition method based on neural network optimized by fractional order ant colony algorithm

Abstract

The invention relates to the field of computer vision and image processing, and discloses a forensic monitoring face recognition method for optimizing a neural network based on a fractional order ant colony algorithm. The training of the neural network is optimized to enable the neural network to quickly converge to global optimum, thereby improving the training efficiency and recognition accuracy of a recognition model, and improving the recognition accuracy of the recognition model. And thus, the face recognition accuracy of court monitoring is improved. The method comprises the following steps: a, acquiring a court trial site monitoring video stream; b, extracting key frames from the monitoring video stream, and performing face detection in each key frame; c, carrying out correction and normalization processing on the detected face image; d, performing feature extraction on the face image processed in the step c; e, recognizing the extracted face features by using the trained face recognition model, and outputting a recognition result, wherein the face recognition model adopts a neural network architecture, and a neural network is optimized by using a fractional order ant colony algorithm during training.

Description

technical field [0001] The invention relates to the fields of computer vision and image processing, in particular to a court monitoring face recognition method based on fractional ant colony algorithm optimization neural network. Background technique [0002] With the advancement of computer science and the high requirements of smart courts, more and more computer theories and methods are applied to it. In the court application scenario, the monitoring of the trial scene and the early warning of abnormal behavior of personnel are of great value. Facial recognition is the basis for on-site monitoring of court trials and early warning of abnormal behavior of personnel. [0003] At present, the mainstream framework of face recognition is divided into four steps: (1) face detection, that is, to detect the existence of faces from court scenes and determine the position of faces in the image; (2) face calibration, to calibrate the face Changes in face scale, illumination and rot...

AI Technical Summary

Problems solved by technology

Although BP neural network can be used to match and recognize facial features, it has the disadvantage of slow learning convergence and easy to fall into local extreme points.

Images