A New Human Detection Based On Rectangle Fittin

Human detection means the computer to detectthe appropriate manikin in the image or video.In recent years,the technique ofhuman detection in pattern recogniti-on and computer vision researchers are increasingly,and it receivied the attention ofscholars.According to the current literature,it seems that more and more human body detection algorithm is proposed,and it play aimportant role.Because of the posture of human body is uncertainty,the complexity of the background,human blocked in the picture background and the limitations exist in the algorithm,Human detection technology meet some difficulties in all aspects,and it did not have a complete algorithm in this area.This algorithm need to keep constantly innovating and be perfected.Now,the human detection technology has been widely used in intelligent video surveillance systems,intelligent transportation systems,intelligent human-computer interaction systems,and content-based image retrieval system.In this paper,it not only summarizes the human body detection but also study the key and difficult questions.It described the application of the human body that detection in various aspects.According boardcard manikin,human body can be representsby rectangles,human bodydivided into ninesections,such as,the trunk,right arm,right lower arm,left arm,left arm,right thigh,right leg,left thigh.After analyzing several segmentation method,this paper proposed a segmentation algorithm that based onthe JSEG algorithm of color image segmentation,this method can substantially Segmented the module of human body in rich color image,and it not only useshort time but high efficiency.JSEG color image segmentation must select the color space first,and then find the segmentation method of color space.This paper presents a human detection algorithm that based on a rectangular fitting,after color segmentation.The color segmentation module isfitted to rectangle.And removingthose modules thatare too smallor too large.The aspect ratio of modulesbe calculated after human body's regional module was detected.The value of this ratio is a new feature in this paper.If this feature meet with the aspect ratio that derived from the statistical model,thatthe human body can be detected.In this paper,it use color segmentation algorithm combine with rectangular fitting to segmentation color image to calculates the new features of aspect ratio.It developed a new human detection algorithm that is fast and efficient.But in someuncertain circu-mstances,the algorithm have a low efficiency and some mistake appeared,it also needs further study.