A New Method of Dynamic Texture Segmentation Based on Optical Flow and Level Set Combination

Abstract

Dynamic texture is a part of the natural scene. Dynamic texture segmentation issue is research hotspot in the field of computer vision and even robot research area. This paper presents a new dynamic texture recognition method based on level set strategy. The level set function evolutes to the boundary of objects according the optical flow feature of moving objects. Because we use the optical flow based terminal condition, the evolution can fast stop to the edge of dynamic texture object. The new method has the characteristic of fast and accuracy. It overcomes the shortcoming of reinitializing and slow evolution speed in the existing methods. The experimental results show that the method can effectively achieve the dynamic texture segmentation of moving objects and it is a very useful method.