The conventional Deep Belief Network (DBN) often utilizes the method of randomly initializing the weights and bias of Restricted Boltzmann Machine(RBM) to initialize the network. Although it could overcome the problems of local optimality and long training time to some extent, it is still difficult to further achieve higher accuracy and better learning efficiency owing to the huge difference between reconstruction and original input resulting from random initialization. In view of the above-mentioned problem, a kind of DBN model based on Sparse Denoising AutoEncoder (SDAE) was proposed. The advantage of the advocated model was the feature extraction by SDAE. Firstly, SDAE was trained, and then, the obtained weights and bias were utilized to initialize DBN. Finally, DBN was trained. Experiments were performed on card game data set of Poker hand and handwriting data sets of MNIST and USPS to verify the performance of the proposed model. In Poker hand data set, compared with the conventional DBN, the error rate of the proposed model is lowered by 46.4%, the accuracy rate and the recall rate are improved by 15.56% and 14.12% respectively. The results exhibit that the proposed method is superior to other existing methods in recognition performance.

Multi-view stereo image composition is largely dependent on the type of multi-view stereo display device. Currently, optical-plate-based multi-view stereo LCD display is most popular, while there is lack of a general composition method for this kind of display. A new general composition method was proposed for the most popular optical-plate-based multi-view stereo LCD display. The method is made up of three parts, i.e. sub-pixel judgment, sub-pixel sub-sampling for each view, and sub-pixel arrangement and composition of each view. This method covers all the possibilities of optical-plate-based multi-view stereo LCD display, with good applicability and popularity. The correctness and validity of the proposed method is verified by experiments.

Based on MPEG-4 FGS technology, a new scalable coding algorithm for multi-view stereo video was proposed, combining FGS scalability with views scalability. In the proposed algorithm, the middle-view sequence was coded as a base layer and a FGS enhancement layer, with other view sequences all predicted from the middle-view sequence yielding each view's FGS enhancement layers. In this coding structure, three cases (I, P, B frames) were considered, and the corresponding improvements had been implemented. The new method can provide a flexible scalability performance, and it can match many different kinds of transmission requirements. Experimental results proved the multi-view scalability performance of the proposed scheme.