Implementation and Analysis of Eidochromatic Transform for Color Image Compression

With increasing use of color images for various applications, transmitting them quickly and with high fidelity has been possibly the greatest challenge today. In order to achieve this objective, the image has to be compressed to the greatest extent with the loss of as little image information as possible. Many current popular color image coding formats enhance the compression by the use of a color component transform, which reduces the redundancy across the three color channels and a subsequent transform, which relates the values of pixels in a single color channel at proximate but different locations. The objective of this thesis work has been to reduce the redundancies in all the color planes using a recent transform which translates into a better compression performance for a color image compression codec. The emphasis has been to optimize the novel Hybrid Vector Scalar Quantizer (HVSQ). To achieve this objective, a spatiochromatic transform called Eidochromatic transform has been studied and implemented. This transform step is performed after the color transform to luminance-chrominance plane and their subsequent wavelet transform. The tests performed have shown that eidochromatic transform concentrates the energy in particular bands in the chrominance channels. Also discussed are classes of images on which this algorithm shows significant Peak Signal to Noise Ratio (PSNR) improvement and some of its limitations.