Aiming at the shortcomings of object recognition based on shape features

a new method composed by moment invariant and geometric features based on contour is proposed.In order to calculate the mean

standard deviation and coefficient of variation of the features

a sufficient number of training samples is gathered.Those features are selected and weighted dynamically according to the coefficient of variation to establish the mixed contour feature vector model of the object.The mixed contour feature vector of the object in the scene is obtained on line

and normalized by Gaussian normalization model.The weighted Euclidean distance between scene object and model object is calculated to recognize object according to the threshold.The experimental results verify the validity and practical value of the method.