Evaluating the effect of MIPM on vehicle detection performance

ABSTRACT

The introduction of new techniques to improve the robustness and accuracy of vehicle detection is always important for the intelligent transportation system as it may face different problems and challenges. Conventional image-based vehicle detection methods have presented difficulties in acquiring good images due to perspective and background noise, poor lighting and weather conditions. We propose a high-accurate, vehicle detection method by using Modified Inverse Perspective Mapping. Thus, the perspective effect is removed, and then the Hough transform was applied to extract road lines and lanes. Gaussian Mixture Model and chromaticity-based strategy were applied to segment the moving vehicles and tackle shadow effects, respectively. We evaluated the performance of the proposed method under recorded videos in Madrid and Tehran (with different weather conditions at urban and interurban areas). Results indicate that the proposed approach is feasible, and more accurate compared to others, especially when facing bad weather conditions and lighting variations in different environments.

KEYWORDS:

• Inverse Perspective Mapping; Image transformation; MIPM; GMM
• Vehicle detection
• Detection accuracy
• Counting
• Bad weather conditions

Disclosure statement

No potential conflict of interest was reported by the authors.