The Research Of Fast Lossless Hyperspectral Image Compression Algorithm

In recent years, with high spectral increasing amount of image data, contradictory limited channel capacity and large amounts of data between sharply prominent. One solution is effective to reduce the amount of data by compressing and accelerate the speed of data transmission. Because the information data received by the ground without any loss, it must be lossless data, and image data needs to be re-encoded and stored on airborne or spaceborne platform, which made a high compression rate requirements. In addition, data compression can also play encrypted, the protective effect of the data. Therefore, hyperspectral image data compression is not only effective, but necessary. This paper presents a fast lossless hyperspectral image comp ression algorithms.For fast hyperspectral image compression algorithm, we need to find a suitable transformation, at low computational complexity premise of performance as well as possible. Lapped biorthogonal transform the idea of using the lapped transform basic removes blocking artifacts and become energy correlation value can not be obtained under the conditions of the optimal concentration of transformation, and its adaptation for the fast algorithm is feasible. Thus, the algorithm selected LBT as a whole core compression algorithm.In this paper, based on 3D-LBT transformation program, all of which can be calculated using only shift and add complete and the computational complexity is very low. Although the wavelet transform is also no blockiness, b ut its operation is much greater than the LBT; although LBT slightly higher than the theoretical average computation DCT, but it has a DCT do not have a high compression ratio and no blockiness two advantages. LBT in achieving the time and can be further simplified in structure theory LBT algorithm, thus further compensate for its computation speed between the DCT disadvantage.For transform coefficients obtained, this paper proposes a 3D coefficient prediction program, 3D prediction coefficient scanning pr ogram, and the two parts and entropy coding combinations constitute an encoder, a 3D transform coefficients for further compression. 3D prediction scheme decided by the energy distribution of transform coefficients determine the best prediction mode, thereby reducing the amount of data to be encoded to the maximum extent possible; 3D scanning solution prediction coefficient prediction coefficient values grouped by size, and the proposed three-dimensional zig-zag scanning mode, so scanned in a manner less coefficient entropy arrangement; the interval arithmetic coding in both speed and compression performance superior to variable-length coding and arithmetic coding, so I chose it as entropy encoding module algorithm.Next, a detailed analysis of the spectral dimension and spatial dimension of hyperspectral images relevant and full use of the image of the correlation coefficient prediction of 3D programs and 3D prediction coefficient scanning program has been improved compression performance slightly improved algorithm enhancement a substantial decline in the computational complexity of the corresponding module. Finally, the simulation experim ents, AVIRIS 2006 Yellowstone calibrated scene 0/3/10 average data compression ratio of 2.5 or more, the compression rate of about 1.38 Mpixel / s.