Research On Pedestrian Detection Algorithm Based On Multi-feature Fusion In Complex Scene

Pedestrian detection is an interdisciplinary research topic,including image processing and pattern recognition.As science and technology advance,pedestrian detection has attracted increasingly number of people's attention.It has a wide application background in the fields of intelligent transportation,Automatic control,human-computer interaction and so on.In the past few years,though it has progressed apparently,it is still very difficult to take account of both the accuracy and the real-time detection because of some factors like the changing posture of pedestrians and the complex backgrounds.First of all,this paper introduces the current research status of pedestrian detection algorithm,analyzes the existing problems and presents frequently used features and classifiers.Besides,the algorithm of pedestrian detection is studied,and the details are as follows:(1)For the deficiency of single feature,the detection method of fusing multiple channel features is proposed in this paper.In the training phase,the Aggregated Channel Features(ACF),the texture features of Local Binary Patterns(LBP)and the contour features of Sketch Tokens(ST)were extracted,and trained separately by the Real Adaboost classifier.In the detection phase,the cascading detection idea was used.The ACF classifier was used to deal with all objects,then the complicated classifier of LBP and ST were used to gradually filter the result of the previous step.The experimental results verify that LBP and ST can be used as a complementation of ACF.So some objects of false detection can be eliminated in the complicated scenes and the accuracy can be improved.At the same time,The efficiency of multi-feature algorithm is ensured by the cascading detection.(2)According to the characteristics of the movement of pedestrians and the correspondence between position and height of pedestrians in images,the pedestrian detection algorithm combining motion feature with location estimation is proposed in this paper.The motion feature and the aggregated channel features(ACF)were extracted,then these features were trained by theclassifier,and an evaluation model was built on the possible position of pedestrians.In the detection stage,the candidate pedestrian areas were determined by the classifier,and then the non maximum suppression algorithm was used to eliminate redundant windows.Finally,the model of position evaluation was used to judge the pedestrian candidate regions to eliminate non-target pedestrian.The experimental results verify that this method can effectively improve the accuracy of the algorithm based on the ACF.