A simplified image compression technique based on Haar Wavelet Transform

The main objective of source coding is to represent the symbols or messages generated from an information source in a suitable form so that the size of the data is reduced. In image compression we use JPEG where huge number of zero is generated in medium and high frequency region of transformed image using the combination of DCT (Discrete Cosine Transform) and quantization. This is done to take the advantage of `run-length coding' to reduce the size of an image. The process is lossy compression but provide good illusion at a glance. In this paper we use Haar wavelet matrix instead of DCT to transform the image into frequency domain. Again instead of weighting matrix of quantization, we use mask of very few 1s in DC and low frequency region to get huge number of 0 in each block of transformed image. The image is recovered using of IDWT (Inverse Discrete Wavelet Transform) and the MSE (mean square error) and SNR (signal to noise ratio) are measured varying the percentage of zero per block or mask.