Image transmission based on neural network approaches

A method for designing the vector quantizer (VQ), called concentric-shell partition vector quantization (CSPVQ), is introduced. Neural networks using the frequency sensitive competitive learning (FSCL) algorithm using the concentric-shell partitioning for VQ is presented. This new technique will show the speed and the simplicity of the neural network while retaining the computational advantages of the image. A high compression ratio is achieved which reduces the channel transmission bandwidth and improves the utilities of the communication system