摘要：The Isometric mapping algorithm is an unsupervised manifold learning algorithm, with no consideration of the class of training samples, while supervised isometric mapping treats the difference among classes equally. Considering the inner relationship between different expressions, we have proposed isometric mapping algorithm based on expression weighted distance, which assigns weighted values according to different sample distance in order to make full use of knowledge of expression classes when calculating the geodesic distance between training samples. We use incremental isometric mapping algorithm on new samples so as to simplify computation significantly when dealing with new samples. Then k-NN classifier is applied to classify different expression features. The facial expression recognition experiments are performed on the JAFFE database and the results show that this proposed algorithm performs better than ISOMAP algorithm and supervised ISOMAP algorithm, and it is more feasible and effective. /mtp_HPc (Envelope Mean Variance method), M2M4 (Second order and Fourth order Moment method)). Furthermore, according to the characteristics of FM modulation system, we put forward the improvement of M2M4 and also discuss the optimal algorithm for underwater acoustic Rayleigh channel, Rician channel and Bellhop channel by comparing these methods.
