The Measure Technology Of Vehicle Distance Based On Machine Vision

Ranging system is an important part of IVS (Intelligent Vehicle system). It has the vital significance to travel's security enhancement and traffic accident's occurrence reduction. Some ranging methods wildly used in present are analyzed in the article. Compared their strong points and shortages, a ranging method based on machine vision is proposed. It adopts monocular CCD vidicon as input device, and processes the input image by computer, finally calculates the distance according to the model on distance between two vehicles.The entire ranging system design is introduced in the paper, including the system structure, the hardware platform, the software simulation platform as well as system's performance, and the ranging algorithm principle and the flow is elaborated with emphasis based on the image processing technologyFirstly, the wavelet theory is used to carry on the image de-noising and the image enhancement. The iterative threshold segmentation arithmetic is proposed to segment the lane and the Krisch operator is selected to detect the image edge. Next, the ranging algorithm's including three important steps is researched deeply. Regarding the lane line detection, a kind of improvement method based on fuzzy theory is proposed. It overcomes the shortcomings of conventional Hough transform method. The experiment results show that the method is effective and it can detect the line accurately. Then, in order to find the vehicle accurate position, the method based on road surface average gradation and the winder energy are presented. Finally, with the aid of the geometric transform theory, the monocular ranging model is established and the leading vehicle distance is calculated. The experiment indicated that this set of algorithm is real—time and accurate enough to utility.