Apple Shape Classification Using Level Set and Motion Estimation

A new algorithm for apple shape classification using level set and motion estimation was proposed. At first, a standard class shape apple images database was construct by expert, and then the level set representations according to signed distance transforms were used, which are a simple, robust, rich and efficient way to represent shapes; second, the unknown shape class apple was aligned to the standard known shape class apple by motion estimation; at last, based on difference area that was represented by level set, a new metric was developed. The unknown class shape was classified to the same class as the standard class shape which metric was minimum. Promising results were obtained on experiments showing the efficiency and accurate of our algorithm.