The Extraction Of Feature Points In Motion Estimation Based On Mesh Model

Motion estimation/Motion compensation (ME/MC) is one of key techniques for removing redundancy in image video compressing. Prompted by the development of still-image compression, wavelet theory has been widely used and many improved algorithms have appeared. Using redundant wavelet is the prominent research direction at present. The development of motion model promote the video coder from another aspect. Irregular mesh has been research pop for its better property of approximate the texture and the profile of the motion object.Extracting points is very important in motion estimation based on irregular mesh. On the background of motion estimation and motion compensation scheme based on irregular mesh , and the situation of extracting feature points with a predefined threshold, this paper studied the method of extracting feature points with adaptive threshold. The Otsu method is employed to get the threshold, and the effectiveness has been proved with a large number of experiments. The choosing of proper control points is completed using a shift invariant redundant wavelet transform, form a correlation mask and extracted feature points according to different thresholds.The images samples are selected from the international standard test warehouse, and are QCIF/CIF(YUV). The luminance component is used in the experiments. The experimental results presented in this paper show that the proposed method can get different threshold according to different image and the points extracted are more exact. In motion estimation, this method can confirm a new I frame and improve the veracity of motion estimation and establish a foundation for the research of motion compensation theory.