Pedestrian identification method based on gradient cascade SVM (Support Vector Machine) classifier

Abstract

The invention discloses a pedestrian identification method based on a gradient cascade SVM (Support Vector Machine) classifier. In the method, pedestrian identification is carried out by use of a risk-sensitive classifier and a voting mechanism, the detection precision is improved to a certain degree; and furthermore, a gradient cascade architecture is judged by use of an assembly line type from coarse grains to fine grains, negative samples which are easily judged can be completely detected in first levels, the classification speed in the detection phase is greatly accelerated, the operation efficiency and the total processing speed are improved. The method can be adapted to pedestrian identification in scenes with a small pedestrian flow and a lot of non-pedestrian targets.

Description

technical field [0001] The invention belongs to the technical field of video monitoring, and relates to an intelligent video monitoring method, in particular to a pedestrian recognition method based on a gradient cascaded SVM classifier, and is especially suitable for pedestrians in a scene where there is less pedestrian traffic and a large number of non-pedestrian areas exist identify. Background technique [0002] Head and shoulders recognition is a kind of pedestrian recognition method, by detecting the head and shoulders features of pedestrians in the video to determine whether they are pedestrians. Relying on image processing technology, this technology is mainly implemented based on pedestrian feature extraction and recognition algorithms. The purpose is to realize the recognition of pedestrians passing through a specific area by processing pedestrian information in video streams. It is an intelligent video surveillance system. an important branch of . So far, the ma...

AI Technical Summary

Problems solved by technology

The SVM classifier has unique advantages for nonlinear small sample and high-dimensional data classification, but the training of SVM requires a large amount of storage space, which will affect the performance of the SVM classifier when there are a large number of non-pedestrian targets in the detected picture

In addition, for the case of linear inseparability, SVM uses a kernel function to map the input space to the feature space for solution, which makes the nonlinear SVM slow in the prediction process, which is not conducive to rapid identification.

[0005] Since the existing methods of pedestrian recognition using HOG features and SVM classifiers all use a single high-dimensional HOG feature and a non-cascaded SVM classifier, when there are a large number of non-pedestrian targets similar to the sky in the video scene, the performance of SVM Classification performance will suffer

Images