| Summary: | 碩士 === 義守大學 === 資訊工程學系 === 88 === Fractal image compression has the advantages of the high retrieved image quality, high compression ratio and resolution independence. But it is time consuming in the encoding phase. In this thesis, one proposes two methods to improve the encoding speed and the compression ratio, which are the fast discrete Hadamard transform (DHT) method and the spatial correlation method.
The baseline method of fractal image compression is utilizing the self-similarity between the sub-image blocks in the same image to achieve the purpose of image compression. Each sub-image block must be compared to the eight iosmetries of all bigger sub-image blocks in the image to find the most similar block. It is spends a lot of time for such full search mechanism. In this thesis, two methods are used to improve the encoding speed.
The first method is to remove the redundant computations in the eight iosmetries using the fast DHT method. When the eight iosmetries are transformed to frequency-like domain by DHT, one can find that there are only differences between some columns and rows in those iosmetries. Depending on this characteristic, the computation of the eight iosmetries can be reduced to two computations. Thus the encoding speed can be improved 2.5 times, and the retrieved image quality is completely the same as that of the baseline method.
The second method is to reduce the searching space of each block to be encoded by using the spatial correlation method. The main concept is utilizing the correlation between the adjacent blocks in an image to reduce the searching space. Since the searching space is reduced, the comparison computations of each block to be encoded will be much less than that of the exhaustive searching method. The encoding speed is about 1.5 ~ 8 times faster than the baseline method and the bit rate is improved about 10% ~ 31%. Since the two methods mentioned above are independent, they are combined as a new fast Fractal Image Compression algorithm. The encoding speed will be improved up to 3 ~ 20 times and the retrieved image quality is almost the same as that of the baseline method.
For the better image quality and bit rate control, the new fast method can be applied to the quadtree decomposition (QD) scheme and the quadtree recomposition (QR) scheme. The encoding speed is improved 2.3 ~ 9.4 times and the bit rate is improved 4% ~ 14% than that of the QD baseline scheme.
|
|---|
