An Adaptive Two-Stage BPNN–DCT Image Compression Technique

Neural Networks offer the potential for providing a novel solution to the problem of data compression by its ability to generate an internal data representation. This network, which is an application of back propagation network, accepts a large amount of image data, compresses it for storage or transmission, and subsequently restores it when desired. A new approach for reducing training time by reconstructing representative vectors has also been proposed. Performance of the network has been evaluated using some standard real world images. Neural networks can be trained to represent certain sets of data. After decomposing an image using the Discrete Cosine Transform (DCT), a two stage neural network may be able to represent the DCT coefficients in less space than the coefficients themselves. After splitting the image and the decomposition using several methods, neural networks were trained to represent the image blocks. By saving the weights and bias of each neuron, by using the Inverse DCT (IDCT) coefficient mechanism an image segment can be approximately recreated. Compression can be achieved using neural networks. Current results have been promising except for the amount of time needed to train a neural network. One method of speeding up code execution is discussed. However, plenty of future research work is available in this area it is shown that the development architecture and training algorithm provide high compression ratio and low distortion while maintaining the ability to generalize and is very robust as well.