| Intelligent Vehicle System (IVS) is a new developing cross discipline thrived recently. Its research involves a great deal of fields, such as computer measurement and control, computer vision, sensor data fusion and vehicle project, etc. We can say that the research of the intelligent vehicle is an integrative application of computer vision and computer control on the vehicle project.When the driver drives a car, the messages he or she received, such as traffic signals, road signs, etc., are mostly obtained by vision. The computer vision system is an important problem to be figured out to realize Collision Avoidance System(CAS) of IVS, Which becomes the focal point studied in recent years too. In order to make the vehicle be able to be controlled automatically only according to the images, which are collected by the vision-sensor, we must extract the feature of the images by some special technique of image processing, for example, segmentation of interested region, finding out the eigencurves or just analysis of the texture. The primary subject of this paper is the algorithm of extracting eigencurves.The distortion often takes place in the picture during the process of collecting and transmission. In order to improve the visual effect of the image, and make people or computer understand it better, we have to enhance the image first.After the process of image enhancement, we begin to extract the feature of the image. The edge is the most important characteristic in the picture. Edge detection is the foundation of eigencurves extraction. There are so many kinds of algorithms on edge detection, among which differential operators are effective tools. According to the results of the experiments, we choose the Log edge operator to detect the edge of the image, because it can give a better edge image even if the image is polluted by noise. In order to realize the CAS, we need to extract the interested eigencurves from the image, so that the computer can control the vehicle according to the location of the curves and make it run in the safety range. We can see that the straight lines are the most important curves in the pictures of highway. They reflect the position of the route directly. So, we regard straight line extraction as the focal point of our research. In addition to meet the speediness request of vehicle control, the algorithm of curves extraction must be real-time. But the traditional algorithms can't meet the real-time request because of theirs storage burden and computational complexity. Accordingly, we propose a modified algorithm in order to improve the efficiency. The Hough Transform(HT) is one of the most often used tools for curve detection. The original HT is commonly used to detect straight lines in a binary image. It is a voting process where each pixel of the original binary image votes for all possible patterns passing through that point. The votes are accumulated in an accumulator array, in which its peaks correspond to line segments. The main advantages of the HT are its robustness to image noise and that it can determine the slope and the distance from the origin (polar parameters) of discontinuous straight lines. The disadvantages of the HT are associated with its large storage and computational requirements. In addition, the information given by the peaks of the accumulator array are only the polar parameters of the straight line and the total number of pixels that belong to it. Unfortunately, the HT does not determine the exact position of each pixel in the straight lines. Accordingly, Inverse Hough Transform(IHT) algorithm has been proposed. This algorithm reconstructs correctly the original image pixel by pixel, using only the data of the Hough space. But this algorithm can't fit the real-time request, too. The IHT algorithm, which ignores the characteristic of the curves, is so "blind". And for this reason, we propose a new fast Inverse Hough Transform based on curve feature (CF-IHT). First, the CF-IHT algorithm picks out some feature points in the image at random...
|
PDF Full Text Request