Vehicle Rear Obstacle Detection Algorithm Design

The steady growth of world economy, as well as vehicle prices fell sharply lead to the continuously rising in privately owned motor vehicle in the recent few years, While we enjoy the convenience brought by the motor vehicle, the increase in traffic accidents is also plagued us. Therefore, how to reduce the traffic accidents and the losses caused by the accidents have became our focus.The thesis is based on road recognition technology and obstacle detection technology, it affords a comprehensive analysis and uses a variety of image pre-processing technology and provides a plentiful environmental status information. And in this thesis, the primary means of accessing to information is the sensor which is installed in an automotive vehicle rear, and this makes the study of moving vehicles rear obstacle detection come true. The thesis implements the Harris corner detection and gives an improvement, it provides a new method which is based on sub-pixel, eliminates the influence affected by the threshold, and makes the test result more accurate. Moreover, it also puts forward a new algorithm called three inter-frame difference method. This method uses neighboring three frames to make different inter-frame difference, then uses the two different inter-frame results to conduct an operation "&", and then runs the binary operation. Finally, the targets are located in the image. This method not only lasts the advantages of the two inter-frame difference method, but also overcomes its disadvantages.The correctness of the three inter-frame difference is proved. And the algorithm is implemented by selecting the open source project OpenCV developed by Intel, and using the integrated development environment VC++. More importantly, the experimental results are satisfactory.