A new Image compression technique using principal component analysis and Huffman coding

Abstract

Principal component analysis (PCA) is one of the most widely used techniques for dimension reduction. It exploits the dependencies among the variables and represents the higher dimensional data in the lower dimensional with more amenable form, without losing a countable information. In this paper, we present a new image compression technique that uses PCA and Huffman coding. The input image is first compressed by using PCA, few of the principal components (PCs) are used to reconstruct the image, while the other less significant PCs are ignored. The reconstructed image is further quantized with dither to reduce contouring, occurred due to less number of PCs are used in image reconstruction. Finally, the Huffman coding is applied on quantized image to remove coding redundancy. The proposed image compression technique is applied on several test images and results are compared with JPEG2000. Comparative analysis and visual results clearly show that the proposed technique performs better than the JPEG2000.