Low bit-rate video coding using image triangulation and warping

We proposed a new video coding method that results in a lower bit-rate than the MPEG-1 method. Instead of dividing the image frames into 8×8 blocks, the frames are triangulated according to a set of control points. The control points and the edges of the triangular blocks are located at the edge features in the image. These control points in the sequence are tracked and their motion is encoded and sent to the decoder. The prediction error is computed as the difference between the original image and the estimated image that is reconstructed by warping the corresponding triangular blocks in the previous frame. The prediction error is encoded using vector quantization (VQ). Experimental results show that the bit-rate needed to encode the Susie sequence is about one-fourth the bit-rate needed in the MPEG-1 method. By properly selecting the number and position of the control points, it is expected that very low bit-rate video coding can be achieved