Human detection tracking and crowd behavior analysis using image processing

Abstract

To examine the human behaviors, a large amount of effort is put in video newlinesurveillance. If a crowded scene is pondered, the attributes that are acquired will help to newlinedetect anomaly with a significant component. However, it is difficult to gather information newlinefrom a complicated video scene as it demands prior details and also involves enormous newlinecosts in performing mathematical computations. newlineThe problem may be addressed using Multi-observational Detection and Tracking newlineApproach (MoDTA) by obtaining the location of people from image and detecting the newlinedefined values at the pointed locations which normally grow according to the number of newlinepeople, thereby calculating the weighted values of independent attributes. Moreover, to newlinesimplify the complex scene tracking, advection molecule is utilized in motion newlinerepresentation. Correlation coefficient is utilized as a function of template detector to newlineevaluate the forthcoming object. Furthermore, to evaluate the system, MoDTA is matched newlinewith other existing tracking and detection methods. newlineIn video processing, identity recognition and complexities due to unusual behavior newlineare well-known problems. Choosing the area of concern and by tracing them for a short newlineinterval of time by a feature detector, anomaly behavior can be enumerated. The detector newlineoutput demonstrates the trade-off among the visual movement and tracking of object as newlineseveral regions display many kinds of moving patterns. By using Distribution Based Crowd newlineAbnormality Detection (DCAD) approach, the detector set presents a spatial and temporal newlinewindow and the object trajectory that passes through these spatial-temporal cubes can be newlinecaptured. Further, magnitude and direction of the trajectories and salient points are encoded newlineby the robust detector. Extent of interaction among the people in the pixel space is obtained. newlineSupport Vector Machine is used for classification. Outliers of the usual situation are the newlineabnormalities. newlineIn the era of computer vision, by virtue of complexities in motion pattern and newlinebackground disorders.