The Appliance Of Wavelet Analysis In The Compression Of Image

It is not occasional that message is changed from analog to digit. Digital signal prevails over simulated one, it has a mighty ability to resist interaction in transfer, its store is very convenient, and its circuits are easily integrated. But a digital image need occupy considerable store units, for example, the numeral rate of a trenchant television signal amounts to 1.3Gb/s. It is apparently difficult and uneconomic to store in real time or transfer so much data capacity. So the compression of image becomes necessary and important.It is the compression of image that removes unnecessary communication of the image and remains useful communication to us according to the relevance of image. Its key technology is how to encode image, meanwhile, the technology also is a core problem in multi-media communication, digital broadcast and television. The compression of image acts an important role on storing and transferring image in a limited store space and definite channel breadth.This article puts emphasis on how to improve compression ratio of image on the premise of ensuring the compressed image with good sight when wavelet analysis is applied to the compression of image. The process of coding image is following: the first, converting image into vectors; second, transforming these vectors into wavelet coefficients; third, quantifying coefficients into symbol stream; at last, compressing symbol stream into bit stream.Wavelet analysis is a method of analyzing the time-dimension of a signal. It has a character of multi-resolution analysis and the ability to detail local characters of a signal in both time and frequency. Its size of transforming window keeps constant, but the shape of both its time window and its frequency window may be changeable. That is to say, the coefficients in low frequency have higher resolution to frequency and lower resolution to time than the ones in high frequency. Conversely, the coefficients in high frequency have higher resolution to time and lower resolution to frequency than the ones in low frequency. After a primitive image is transformed in wavelet, almost its energy is concentrated in the low frequency coefficients, minority is scattered in the high frequency coefficients. In other words, the low frequency coefficients reflect the "crude portion" of a primitive image, the high frequency coefficients reflect its "minute portion". Of course, it isimpossible to improve the compression ratio by wavelet, but omitting high frequency coefficients and remaining low frequency coefficients will not only greatly improve the compression ratio, also enhance the image's ability to resist interaction.It is not enough to only apply wavelet analysis for utmost improving compression ratio in the premise of keeping compressed image good sight. For this destination, it also need rely on some coding technologies. SOFM arithmetic methodology, LBG arithmetic methodology, DPCM encoding step by step, wavelet zero tree compression etc, are common methods applied in the encoding of image at present.It is impossible for me to make a great break through and innovation because of shortage of time in understanding wavelet and image compression. My principal work is these: transforming image in wavelet at first, then encoding wavelet coefficients in SOFM arithmetic methodology, LBG arithmetic methodology, DPCM encoding step by step, wavelet zero tree compression separately, reconstructing the image at last, so researching how to improve compression ratio, hi addition, the article puts forward a new arithmetic methodology, SL arithmetic methodology, on the base of comparing SOFM arithmetic methodology with LBG arithmetic methodologyWhat it needs emphasis is the variety of reconstructed image because of the diversity of mother wavelet. There isn't a conclusion that which of wavelets is suitable for which type of image. This is why people are keen on researching wavelet analysis. All the experiments are made in a wavelet, "db1", in the article. In addition, the technical parameters in SOFM arith...