An Image Processing Method for Intelligent Vehicle based on MT9V034.

In order to comply with the development trend of intelligent vehicles, an image processing method was designed for intelligent vehicle. The intelligent vehicle used TC264DA as the core control unit and acquired track images through MT9V034 camera. After image pre-processing such as down-sampling, binarization and noise reduction, this project used eight-neighborhood boundary tracing algorithm to obtain the track boundary and the seed area growth algorithm to complete the track. Next, we carried out some tests on a specific track and the results showed that the recognition rate of cross elements can still reach 95% at the speed of 2.8. This speed is competitive in national smart car competition. An improved algorithm based on Otsu algorithm was proposed. It divides the image into bright and grey parts and calculates the segmentation threshold for each part separately. Tests show that the algorithm significantly improves the binarisation of images that are unevenly illuminated.