Wavelets and Neural Networks based Hybrid Image Compression Scheme
Sai Kiran Arcot Ramesh · 2013
Images having large data quantity needs more space for storage and very high data rates for transmission. As a result, highly efficient image compression methods are under wide attention. Employing more than one traditional image compression algorithms, results in hybrid image compression techniques. This paper aims at implementing a wavelet transform and neural network based model for image compression which combines the advantages of both wavelet transformations and neural networks. Images are decomposed using Haar wavelet filters into a set of sub bands with different resolutions corresponding to different frequency bands. Scalar quantization and Huffman coding schemes are used for compression of different sub bands based on their statistical properties. The coefficients in low frequency band are compressed by Differential Pulse Code Modulation (DPCM) and the coefficients in higher frequency bands are compressed using neural networks. Using this scheme we can achieve satisfactory reconstructed images with increased bit rates, large Peak Signal to Noise Ratio (PSNR) values. Image compression using cosine transform results in blocking artifacts, which can be eliminated using wavelet transforms, on the other hand neural networks reduce Mean Square Error (MSE). Empirical analysis and calculation of required metrics is performed.