An Environment Recognition Algorithm Based on Weighted Cloud Classifier

Environment recognition is a necessary prerequisite for behavioral decision and intelligent control of intelligent vehicle. In order to improve the surrounding environment recognition ability of intelligent vehicles, a weighted cloud classifier is constructed to recognize the content of image captured by vehicle camera. By researching the image texture characters, several cloud classifiers are trained based on the first and second order statistical characteristics of texture to identify the image content preliminarily. Then the cloud models are combined with adaboost algorithm to construct a weighted cloud classifier. Experiment results show that through weight optimization, the weighted classifier can realize multi-target recognition, and achieve good results in the recognition of vehicle, road, lane line and other targets. The weighted cloud classifier will play an important role in improving the environment recognition and behavioral decision capacity of intelligent vehicle.