Improvement Algorithm Of Spiht Of Image Compression Based On Wavelet Transform

With the development of the computer and the network technique ,people need the high quality , appropriate size of the image .We must store and transmit the big image in the limited space and bandwidth .We can get the different image based on our need . It requests that the compression technique of the image has both good compression efficiency and flexible compression code ratio .Some traditional Compression Algorithm of the image can' t meet the request .Algorithm of wavelet transform has some good quality : excessive differentiate , focus of energy , fast operation of decompose and compose . It is used in many fields of image compression .The page first introduces analyze theory of the wavelet . Then it brings up the fast algorithm of Mallat based on wavelet transform and excessive differentiate .Wavelet transform on the fields of image compression needs solving some questions . The first is the choice of wavelet basement . It is very complicated , from the direction of flatness and the length of filter we usually choice double wavelet basement;the boundary of the image is limited ,we must extend it to get good messages;the more ,many ways can estimate wavelet coefficient, we choice the zero tree to get it;finally, we usually choice three or four of the wavelet grade .The page researches embedded coding . On the foundation of embedded coding we research ZEW algorithm and SPIHT algorithm and find their characteristic . We know SPIHT algorithm is the improvement of ZEW algorithm . SPIHT algorithm constructs two different kinds of zero trees : D(i,j) , L(i,j) .So it can make use of fall of wavelet coefficient . But it has some shortcomings .For the shortcomings of SPIHT algorithm , the characteristic of image compression , vision characteristic of our eyes ,we introduce three improvement ways .The first , SPIHT algorithm has its own shortcomings , we can use the lowest frequency band to initialize nonsignificant coefficient list .so the depth becomes longer in space trees . It can get more zero trees and deal simple .The second , wavelet decompose of the image can get the low frequency band of more energy , it is important;the high frequency band is unimportant . To give prominence to the importance we can deal with the high frequency band firstly . So it can transfer more important messages . On the other way , our eyes are sensitive to the low frequency band , it can enhance image quality .The finally , we use (9, 7) filter to decompose the image , we can find thatthe coefficient of the low frequency band is mostly positive and big .the negative is small . So we don' t transfer the sign of the coefficient in the low frequency .we think they are positive . So we can transfer more bit in the same instance , the compression efficiency becomes high .The experiment tests that the improvement has its own effect . It has good compression efficiency , on the same time we can deal with code bits flexible .On the real realize of the program we explain the program that deals with binary I/O of the image files and PGM files .The finally , we explain that the lifting wavelet and integer wavelet transform can compress the image .