An Improved Algorithm for Image Compression Using Geometric Image Approximation

Our dependence on digital media continues to grow and therefore finding competent ways of storing and conveying these large amounts of data has become a major concern. The technique of image compression has then become very essential and highly applicable. In this paper, the performance of an efficient image coding method based on Geometric Wavelets that divides the desired image using a recursive procedure for image coding is explored. The objective of the work is to optimize the performance of geometric wavelet based image coding scheme and to suggest a method to reduce the time complexity of the algorithm. We have used the polar coordinate form of the straight line in the BSP scheme for partitioning the image domain. A novel pruning algorithm is tried to optimize the rate distortion curve and achieve the desired bit rate. The algorithm is also implemented with the concept of no tiling and its effect in PSNR and computation time is explored. The enhanced results show a gain of 2.24 dB over the EZW algorithm and 1.4 dB over the SPIHT algorithm at the bit-rate 0.0625 bpp for the Lena test image. Image tiling is found to reduce considerably the computational complexity and in turn the time complexity of the algorithm without affecting its coding efficiency. The algorithm offers remarkable results in terms of PSNR compared to existing techniques. KeywordsBit rate, Geometric wavelets, Image tiling, Rate distortion.