Object Detection and Recognition Based on Color and Shape Features

A method of similar objects recognition in complex environment based on color and shape features is proposed. The distributing color space of certain object is obtained by learning-extending method in order to deal with the image captured under the varied lightness. Shape features of certain object with shadow on it or visualizing partly are detected by improved Hough transform. An object is fuzzy judged based on fusing its color belief and shape belief. The validity of this method is proved by the experiments on mobile robot.