Research On Roi Coding Of 3D Volumetric Images Based On 3D TCE Algorithm

With the improvement of various types of hyperspectral technology and medical imaging technology, three-dimensional images are applied widely. The three-dimensional images genetated by those devices have strong correlation between spectrally and spatially. They provide more detailed information, but also product vast amounts of data and bring great difficult to the storage, transmission and application. In addition, in the practical application of the image, people tend to focus on the region of interest frequently and hope that the region have higher reconstructed quality. Therefore, the study of three-dimensional image compression coding of region of interest has become the hot spots of image coding area.This paper presents a ROI compression coding algorithm of three-dimensional images based on 3D TCE. The coding system in this paper first adopts three-dimensional wavelet packet transform to decorrelate hyperspectral image spatially and spectrally. Secondly, transformed coefficients of ROI are handled. Finally, 3D TCE algorithm codes bitplane. In the section on wavelet transform, it researches the compression algorithm of the region of interest of hyperspectral images and three-dimensional medical images based on two different transforming forms of floating-point distortion based on floating-point and integer distortion based on integer. In the part of coefficient processing of ROI, two methods are studied separately. On one hand, the bitplane of transformed coefficients of ROI are lifted in order to improve the importance of coefficients of ROI and give priority to transfer. On the other hand, it is object-based coding. The proposed algorithm is compared with other outstanding algorithm on compression performance, and the experimental results show that in the most bit-rate our algorithm provides better compression performance both of total image and ROI than a variety of excellent embedded wavelet transform coding algorithms.