A low complexity MP3 decoder based on Broadcom embedded platform was proposed. C code level optimization algorithms on inverse quantization, stereo decoding and alias reduction based on PC were proposed to further reduce the amount of memory usage and the computational complex ity. Furthermore, the executable file of the optimized MP3 decoder was generated under the Linux environment, and transplanted to the set top box based on Broadcom embedded platform. Experi ment results showed that the total time for decoding was reduced on the embedded platform, and the goal of real time and fluent playing of audio files was fulfilled, which demonstrated the effectiveness of the proposed MP3 decoder. The proposed MP3 decoder could be applied in fields such xs the set top box based on Broadcom embedded platform and other portable devices.
A fast algorithm based on direction in intra frame downsizing in H.264 is proposed,which used modes information of macroblocks before transcoding and the direction relation of modes between decoding and re-encoding in transcoding.This algorithm also made use of statistics between decoded modes and re-encoded modes,which came from a lot of sequences data experiments.Without full modes encoding,it can improve the speed of reducing intra-prediction frame resolution obviously.Comparing to traditional transcoding,it only needs to compute one of thirteen modes in re-encoding.The experiments show that this algorithm can significantly speed up 92 percent transcoding time in intra-prediction frame of H.264 with slight PSNR degradation.It also can support an improvement in real-time for transcoding and ability of bandwidths changing.
An analytical queuing model is proposed for the classified services of WiMAX network. Simulation model is also developed that corresponds to the Markovian analytical model using Java modeling tool （JMT）. This is a new and efficient discrete event tool for queuing network modeling and workload analysis. QoS metrics have been evaluated for the multi-rate traffic in multiple scenari- os. Results obtained from simulation are compared for validation and analysis. Outcomes show that the proposed model is more efficient than the conventional method by improving residence time, re- sponse time, increasing system throughput and efficiency at queuing level with a slight degradation in call acceptance factor.
A new method of artificial intelligence based on a new improved back propagation neural network （BPNN） algorithm is partially applied in the problem of image restoration. In order to over- come the inherited issues in conventional back propagation algorithm i.e. slow convergence rate, longer training time, hard to achieve global minima etc. , different methods have been used including the introduction of dynamic learning rate and dynamic momentum coefficient etc. With the passage of time different techniques has been used to improve the dynamicity of these coefficients. The meth- od applied in this paper improves the effect of learning coefficient η by using a new way to modify the value dynamically during learning process. The experimental results show that this helps in im- proving the efficiency overall both in visual effect and quality analysis.
A novel supervised manifold learning method was proposed to realize high accuracy face recognition under varying illuminant conditions. The proposed method, named illuminant locality preserving projections （ILPP）, exploited illuminant directions to alleviate the effect of illumination variations on face recognition. The face images were first projected into low dimensional subspace, Then the ILPP translated the face images along specific direction to reduce lighting variations in the face. The ILPP reduced the distance between face images of the same class, while increase the dis tance between face images of different classes. This proposed method was derived from the locality preserving projections （LPP） methods, and was designed to handle face images with various illumi nations. It preserved the face image＇ s local structure in low dimensional subspace. The ILPP meth od was compared with LPP and discriminant locality preserving projections （DLPP）, based on the YaleB face database. Experimental results showed the effectiveness of the proposed algorithm on the face recognition with various illuminations.
