Abstract: This study demonstrates a lower dimension multiresolution and facial expression analysis of facial images using wavelet transform and image decimation algorithm. It minimizes heavy computational load, reduce noise, produce a representation in low frequency domain and hence make the facial images less sensitive to facial expressions and small occlusions. An improved recognition rate is achieved through effective image pre processing and novel feature extraction technique. Within class varying facial expressions effects have been minimized by using image decimation. Novel feature extraction methodology has been used to extract the most suitable feature vectors required for recognition. Experiments on ORL, YALE, FERET and EME color datasets have been performed with success rate up to 99.25%. Model has been also tested on CMU AMP face expression and dataset to evaluate the ability of wavelets and decimation algorithm for varying expression compensation. Hundred percent recognition rate on this dataset is achieved.