A Fast-Based Method for Crowd Abnormal Behavior Recognition

Abstract

The invention belongs to the field of computer vision, and discloses a method for identifying crowd abnormal behavior based on FAST, including: converting video stream images into picture data; performing enhanced preprocessing on images; establishing a mixed Gaussian background model; and performing FAST corner detection Calculate the corner point covariance matrix, obtain the crowd area change curve S according to the value I of the matrix determinant; input the eigenvector formed by the slope value corresponding to the I value on each I value and the curve S in the support vector machine, and carry out Crowd behavior analysis and model training get the crowd behavior prediction value P; according to the P value, the crowd behavior results are obtained, and the abnormal behavior of the crowd is classified and identified. Aiming at the deficiency of the traditional method, the present invention regards the characteristics of crowd corners as a whole to study the situation of different crowds, establishes a crowd behavior model through the calculation of the covariance matrix, and obtains the behaviors of different crowds. It can be used in security monitoring, resource management and other fields.

Description

technical field [0001] The invention belongs to the field of computer vision, and relates to a method for identifying crowd abnormal behavior based on FAST (features from accelerated segment test, accelerated segmentation detection features). Background technique [0002] With the rapid development of economy and technology, tourist attractions and public transportation systems such as railway stations and subway stations often have peak flow of people, and high-density crowds have brought great hidden dangers to traffic safety. Therefore, it is particularly important to monitor crowds, identify abnormal behaviors of crowds, and take appropriate safety measures to eliminate hidden dangers of accidents. [0003] Normally, the analysis of abnormal behaviors targeting people includes two aspects: one is the analysis of abnormal behaviors of target individuals or a small number of targets; the other is the analysis of abnormal events in groups with a large number of people. [...

AI Technical Summary

Problems solved by technology

[0009] Aiming at the problems existing in the prior art that high-density crowds are difficult to segment and are affected by complex backgrounds, various noises, and lighting, the present invention proposes a method for identifying abnormal behaviors of crowds based on FAST. Accuracy, fast crowd behavior recognition using the covariance matrix of corner features

Images