Multi-feature fusion pedestrian detection method based on HIKSVM classifier

Abstract

The present invention relates to a multi-feature fusion pedestrian detection method based on an HIKSVM classifier. The method concretely comprises the following steps of: determining a behavior area in a pedestrian image sequence; extracting multi-item features in the behavior area, and forming fusion features; and employing a trained HIKSVM classifier based on the fusion features to perform pedestrian detection. The method provided by the invention improves a feature extraction method, after HOG features are extracted, a gradient field is changed to an optical flow field through analogy, andHOF features are extracted, shape information and motion information of an object are retained, so that computation complexity of describing a full-size moving object by employing the optical flow field is avoided and the robustness is high; an LBP algorithm is improved to extract texture features through adoption of an LQC algorithm, so that the computation complexity is reduced on the basis without losing texture information; and a histogram intersection kernel (HIK) and an intersection coordinate drop method are employed to perform classification training of the SVM classifier, so that thecomputation complexity is low and the detection precision is high.

Description

technical field [0001] The invention relates to the technical fields of computer vision and pattern recognition, in particular to a pedestrian detection method based on multi-feature fusion of a HIKSVM classifier. Background technique [0002] Pedestrian detection is a challenging task due to their high variability in appearance and wide variety of poses. A robust feature set is crucial for correctly distinguishing pedestrians if one wants to correctly detect humans in cluttered backgrounds or under difficult lighting conditions. [0003] For the problem of pedestrian detection, many descriptors for feature extraction have been proposed. Among them, the Histogram of Oriented Gradients (HOG) descriptor provides superior performance relative to other feature sets, including wavelets. However, the gradient direction histogram represents the edge information of the image, and retains the shape information of the target. It is aimed at a single image, but pedestrian detection i...

AI Technical Summary

Problems solved by technology

[0006] In view of the above analysis, the present invention aims to provide a pedestrian detection method based on the multi-feature fusion of the HIKSVM classifier to solve the problems of low detection accuracy and poor real-time detection of existing pedestrian detection methods

Images