Detecting pedestrians on a Movement Feature Space

PATTERN RECOGNITION

卷:47

Article

Negri, Pablo^1,3  Goussies, Norberto^1,4  Lotito, Pablo^1,2,5 

[1] Consejo Nacl Invest Cient & Tecn, RA-1033 Buenos Aires, DF, Argentina

[2] Consejo Nacl Invest Cient & Tecn, PLADEMA, RA-1033 Buenos Aires, DF, Argentina

[3] UADE, Inst Tecnol, Buenos Aires, DF, Argentina

[4] DC UBA, Buenos Aires, DF, Argentina

[5] PLADEMA UNCPBA, Tandil, Argentina

关键词: Pedestrian detection;    Movement Feature Space;    Histograms of oriented level lines;    Adaboost cascade;    Linear SVM;   

DOI  :  10.1016/j.patcog.2013.05.020

来源: Elsevier

【 摘 要 】

This work aims at detecting pedestrians in surveillance video sequences. A pre-processing step detects motion regions on the image using a scene background model based on level lines, which generates a Movement Feature Space, and a family of oriented histogram descriptors. A cascade of boosted classifiers generates pedestrian hypotheses using this feature space. Then, a linear Support Vector Machine validates the hypotheses that are likeliest to contain a person. The combination of the three detection phases reduces false positives, preserving the majority of pedestrians. The system tests conducted in our dataset, which contain low-resolution pedestrians, achieved a maximum performance of 25.5% miss rate with a rate of 10(-1) false positives per image. This value is comparable to the best detection values for this kind of images. In addition, the processing time is between 2 and 6 fps on 640 x 480 pixel captures. This is therefore a fast and reliable pedestrian detector. (C) 2013 Elsevier Ltd. All rights reserved.

【 授权许可】

Free