A Proficient Method for Traffic Monitoring System
Video processing is a research topic in the area of computer vision. Traffic monitoring is a key issue to be addressed in day-to-day’s life. To address this issue, this system is going to process the video obtained in terms of frame sequence. The system proposed is going to calculate the signal time dynamically by taking into account the number of vehicles present in the current lane. The number of vehicles is counted by segmenting the input traffic data set into frames and comparing each incoming new frame with that of the Dynamic-Background. Dynamic-Background is constantly updated from Dynamic Environment Register (DER) where changes in the background are constantly updated. A counter and a weight value are assigned to each pixel of the Dynamic-Background. For a pixel position in Dynamic-Background the intensity value with the largest count in all frames is assigned. From the Dynamic- Background, an absolute difference of incoming frames of the dataset is taken to obtain the Object of Interest (OOI). From the object obtained, features are extracted using a holistic classifier and those features are compared with that of the existing object’s features to identify the presence of partial occlusion. Depending upon the score for occlusion and the score for objects feature found, the presence of an object can be found. In this paper performance evaluation is done based on the number of objects found in a particular frame. This OOI can be extended to other object counting applications like pedestrian detection, etc.
A. R. Revathi, Dhananjay Kumar