Multi-layered Image Coding Based On Modified FRIT

From the Fourier transform to the wavelet transform, then to the current popular multi-scale geometric analysis, the wavelet transform and the multiscale geometric transforms had a profound impact on many fields as the development of the image processing to be mature and improved. The Wavelet has a strong analysis ability of local time domain in dealing with the singularity of one-dimensional signals. The performance is better than the Fourier transform. But the wavelet which only can process the vertical, horizontal and diagonal directions in dealing with two-dimensional image limits the effective of processing the image edge. To overcome the lack of the wavelet transform, the geometric e rapidly developed which provides bandlet, the curvlet, the ridgelet and other transformation methods. It has become an effective tool in processing two-dimensional image.This main purpose of this paper is to study the ridgelet, and using the ridge transform to the image compression. The Ridgelet transform which is the most effective two-dimensional image processing tools has a great advantage on processing of images.Pointed at the straight-edge-rich images, this paper proposed a new kind of layered encoding algorithm. At first, the image is compressed by the direct valve value coding algorithm, obtained an approximate image R1, then subtracted R1 from R, obtained a detail image R2, according to the characteristic of the image frat coefficients. This paper proposed a modified FRIT algorithm. The set partitioning in hierarchical tress algorithm is also modified to realize the coding of the transform coefficients according to the finite ridgelet transform coefficients. Experiment results show that the proposed algorithm is more efficient than the spiht and could reconstruct the straight edges better.