Pedestrian detection method

Abstract

The invention discloses a pedestrian detection method which comprises the following steps of: obtaining a candidate detection window in the current-frame image by a human body detection method based on HOG characteristics; then determining the corresponding relationship between the current candidate detection window and the human body detection window in the previous-frame image, and standardizing the size of the current candidate detection window; determining to adopt an image division method based on shape prior or an image division method based on the combination of motion information and shape prior for the current standardized candidate detection window according to whether correspondence is established so as to obtain an object division mask of the current standardized candidate detection window; and finally, verifying whether the current candidate detection window is detected wrongly by a pedestrian classifier jointly trained by the object division mask and HOG detection score. Since the motion information and shape prior are integrated into the image division method, the accuracy of object division is improved, and the detection judgment of the candidate detection window is further improved, thus the error detection rate is effectively reduced, and the detection accuracy is improved.

Description

technical field [0001] The invention relates to a pedestrian detection technology, in particular to a pedestrian detection method in a video. Background technique [0002] Pedestrian detection technology has a wide range of applications in visual surveillance, automotive autonomous driving, robot vision, content-based image and video indexing, and advanced human-computer interaction. In practical applications, due to the complexity of the scene, the change of perspective and scale, the diversity of human posture and clothing, and partial occlusion, pedestrian detection faces great challenges. [0003] In recent years, with the deepening of research, researchers have proposed various pedestrian detection methods. Dalal and Triggs et al. proposed a human detection method based on gradient orientation histograms (Dalal N, Triggs B.Histograms of oriented gradient for human detection.In: IEEE Conference on Computer Vision and Pattern Recognition, Dalal N, Triggs B. Human detect...

AI Technical Summary

Problems solved by technology

The above-mentioned pedestrian detection method combining appearance features and motion features enhances the discriminative characteristics of pedestrian detection features, but the extraction of motion features can only be limited to fixed scenes; on the other hand, due to the requirement of continuous multi-frame image Within the detection window, pedestrians cannot move outside the detection window, and the detection window has a fixed position and size in the multi-frame image sequence. Therefore, only small-sized pedestrian detection can be completed, which limits the application environment of pedestrian detection.

Tian et al proposed a timing analysis method based on color and space information (Tian G, QiFH. Feature transformation and SVM based hierarchyI pedestrian detection with a monocular moving camera. ACTA ELECTRONICA SINICA, Tian G, Qi FH. Hierarchical Pedestrian Detection Based on Feature Transformation and SVM under Conditions. Acta Electronica Sinica), this method first uses the appearance detector (Adaboost classifier based on Haar-like features) to preliminarily distinguish the pedestrian detection frame in a single frame image, and then according to the The similarity between the color of the detection frame and the spatial information determines whether a detection frame appears multiple times, and uses the multiple occurrences of the detection frame as the real pedestrian detection result to eliminate false detections. However, this method can only eliminate part of the apparent The occasional false detection frames in the detectors cannot eliminate the continuous false detections in the apparent detectors, so it does not really make up for the lack of pedestrian detection capabilities of the corresponding apparent detectors

[0006] Through the above analysis, it can be found that although the above-mentioned various pedestrian detection methods have achieved good detection results, they still need to be further improved in terms of reducing false positives, detection and positioning accuracy, and detection efficiency.

Images