Vehicle classification technique for automated road traffic census

Bong, Shuk Hui (2014) Vehicle classification technique for automated road traffic census. [Final Year Project Report] (Unpublished)

	PDF Bong Shuk Hui(24pgs).pdf Download (7MB)
	PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3942/3933) Bong Shuk Hui(ft).pdf Restricted to Registered users only Download (19MB)

Abstract

This thesis proposes the development of vehicle classification technique for automated road traffic census prototype to replace manually vehicle classification by applying morphological techniques for image classification. The developed prototype consists of three main phases: the first phase is video frame preprocessing by applying thresholding, masking, image differencing, and median filter to obtain a resultant image for vehicle detection algorithm. The second phase is vehicle detection algorithm by processing the resultant image with 'Sobel · edge detection, first level dilation, binary filling of holes, boundary vehicle elimination, second level binary dilation, and morphological binary open to obtain a final output image for vehicle classification. Vehicle classification is taken into place by drawing a bounding box on the blob in binary image. Blob analysis will be performed to calculate the area of pixels of the vehicle 's blob. The vehicle will be classified based on the experimental area to classify vehicle into three categories which are mini car, saloon car, and bus. The performance of the prototype is tested on three pre-recorded video under homogenous road environment. The experiments results shown that the prototype can achieve an overall accuracy of 88. 42% for vehicle classification. The result of the vehicle classification is affected by the result of vehicle detection. The vehicle will miss or wrongly classified if the vehicle blob does not fall in the region of interest.

Actions (For repository members only: login required)

View Item