A Synthetic Region Selection Strategy for Texture Synthesis Based Video Coding

To save bit-rate, texture synthesis is employed in video coding to generate similar texture instead of encoding the texture region. Without mean squared error (MSE) as the distortion criterion, high quality but different texture can be synthesized with few bits. One key problem that embarrasses the application of texture synthesis in video coding is how to merge the synthesis technique into the nowadays' I, P, B frame based codec, or more specifically, how to perform synthesis cross I-frame. In this paper, we describe a new synthetic region selection strategy which can achieve smooth transition between the synthetic frames and the adjacent non-synthetic frames. In our method, I-frame is still coded using MSE as its distortion criterion. The area of synthetic region should increase or decrease depending on the distance between the current frame and the nearest I-frame. The nearer is the distance from the I-frame, the smaller is the synthetic region in the current frame. Experimental results show a smooth transition is achieved using our selection strategy with great bit-rate saving.