Fast backpropagation neural network for VQ-image compression

Abstract EN

The problem inherent to any digital image is the large amount of bandwidth required for transmission or storage.

This has driven the research area of image compression to develop algorithms that compress images to lower data rates with better quality.

Artificial neural networks are becoming very attractive in image processing where high computational performance and parallel architectures are required.

In this work, a three layered backpropagation neural network (BPNN) is designed to compress images using vector quantization technique (VQ).

The results coming out from the hidden layer represent the codebook used in vector quantization, therefore this is a new method to generate VQ-codebook.

Fast algorithm for backpropagation called (FBP) is built and tested on the designed BPNN.

Results show that for the same compression ratio and signal to noise ratio as compared with the ordinary backpropagation algorithm, FBP can speed up the neural system by more than 50.

This system is used for both compression/decompression of any image.

The fast backpropagation (FBP) neural network algorithm was used for training the designed BPNN.

The efficiency of the designed BPNN comes from reducing the chance of error occurring during the compressed image transmission through analog channel (BPNN can be used for enhancing any noisy compressed image that had already been corrupted during transmission through analog channel).

The simulation of the BPNN image compression system is performed using the Borland C++ Ver 3.5 programming language.

The compression system has been applied on the well known images such as Lena, Carena, and Car images, and also deals with BMP graphic format images.