An extensive study on video based traffic surviellance system in terms of vehicle detection

Abstract

The use of vehicles for transport is rapidly increasing with improvement in the quality of life. However, such an increase in the use of vehicles compounds traffic problems. Therefore, intelligent transportation systems have become very popular research area where more amount of works are done nowadays. Several researchers have been developed for traffic surveillance over the past few decades. The major work of the proposed system consists of five major phases such as background subtraction, noise removal, feature extraction and selection, clustering, vehicle tracking and classification. Initially, a modified background-updating procedure is introduced to smoothly update the background image during congestion. During the background subtraction, models are applied to background intensities to overcome small changes in environment. Vehicle tracking stages have been carried out in two stages: In the first stage, extraction of important features such as symmetry, edge, headlight, brightness, and appearance, day and night time obtained from Improved Particle Swarm Optimization (IPSO) algorithm is implemented that reduces dimensionality of features through Hybrid Principal Component Analysis (HPCA) and during second stage Fuzzy Hybrid Information Inference Mechanism (FHIIM) is implemented for vehicle tracking. Vehicle detection stage, proposed system detects candidate s vehicles results from background subtraction results. In the second phase, stretchy and effectual video grounded traffic surveillance scheme is presented which can observe and track the movement more precisely. This is implemented using a new method termed Hybridized Artificial Bee Colony Method and Independent Component Analysis (HABC-ICA) technique, which is used to improve traffic surveillance scheme newline