Video compression standards follow certain principles for frame identification based on Group of Pictures (GOP). It was observed that instead of following the pre-coded GOP, if frames are adaptively chosen based on simple matching criteria, further processing could be effectively done. It was also noted from the earlier techniques that the presence of B-frames leads to buffering of the memory within the past and future frames during the coding stage. The computational time increases with more number of B-frames. Hence, a novel framework has been emphasized in the present investigation firstly for adaptive frame identification and secondly, replacing B-frames with either I-frames or P-frames using NSEW Affine Translation for the purpose of video compression. The result shows that the performance objective evaluation coincides with the subjective evaluation