Preferential interframe coding for videophones and teleconferencing using fuzzy logic

We present an interframe coding scheme which utilizes fuzzy inference methods to identify image blocks for preferential coding. In applications like videotelephony and teleconferencing the image fidelity need only satisfy the requirements of the human visual system. Therefore different regions of an image can be coded to different accuracies. A good selection criterion is critical to the success of this technique. The fuzzy inference selection criteria presented in this paper uses membership functions defined for inputs such as block location, block interframe difference magnitudes, block interframe difference history and block statistical properties and computes the degree of membership of each block to a given region. Results showing the blocks selected for preferential coding by this method are presented.