Search

In dimensionality reduction, Isometric Mapping (Isomap) is a classical method with non-linear feature transform, but relies on minimum matrix distance function and assumptions. Maximization of Mutual Information (MMI) derives the effective dimensionality reduction transform from the Information Theory, but difficult to get the solution. We present a new method (ISO-MMI) for learning discriminative feature transforms, using mutual information between objective function and transformed feature, based on the Isomap algorithms, by complementary combination of these two methods. Numerous experiments on different data sets comparing with PCA, LDA, and Isomap, show the effectiveness of this proposed algorithm.

Multiple description coding (MDC) algorithms based on the lower-tree wavelet (LTW) are presented and studied, suitable for real-time applications of video delivery. One is based on different quantizers to the wavelet coefficients according to the quardtrees. And the other is based on the structure unanimity of the two descriptions generated from wavelet coefficients according to the bit plane. Experimental results for the transmission of image demonstrate the efficiency of the proposed method.

The dimensionality of face image is high very much. It has a lot of difficulty in face recognition. In this paper, first, the nonlinear dimensionality reduction algorithm Isomap is applied in face image. On this basis, the concepts of vector membership function and membership degree in fuzzy mathematics are introduced, it is presented that the fuzzy matching for a nonlinear function between input and output can be realized by using three rulers (two point rulers and one slope ruler). The affinity between two memberships can be used for assessment to the linearity of the matched curve. The polynomial fuzzy matching based on three rulers algorithm is applied in face recognition. Experimental results show that the recognition algorithm is feasible and has good recognition capability.

A novel image encoding algorithm is presented based on lower tree [3]to solve the problems such as low speed, huge capacity, repeated calculation in advanced compression codings[1-3]. It is an efficient method of grouping the coefficients, coding them and reconstructing them based on the correlation of the wavelet coefficients in multiresolution. And an improved algorithm is proposed. The experimental result shows that the improved algorithm increases the coding speed, reduces the memory and improves the image recovery quality, so it is an efficient method for image encoding.