Low Bit Rate Image Coding Based On Wavelet Transform

Optical imaging systems have widely application in the spaceflight and national defence domains. As the resolution of imaging systems increased, the image data obtained by imaging systems are more and more large. So the research on coding of the image data obtained by imaging systems is more and more important. Two dimensional image coding and three dimensional hyperspectral image coding have become novel research topics in the optical domain. Two dimensional image coding is an important basis of the research on three dimensional hyperspectral image coding. This dissertation mainly focuses on the low bit rate image coding including wavelet transform and embedded bitplane coding, and mainly solves the problems such as complexity, memory requirement, coding performance and error resilient performance. The main work of this dissertation is summarized as follows:First, by expanding the SCLA9/7 with the symmetric factorization of biorthogonal wavelet, the spatial combinative lifting framework and the reversible integer lifting framework of the two dimensional matrix are proposed, and two groups of lifting parameters named 9/7-3 and 9/7-4 which have low calculations are also proposed. The multiplication operations and memory requirement are effectively decreased by combining the spatial combinative lifting algorithm and the overlapped block transform together.Second, the significance tree is constructed for compact description of wavelet coefficients, and four coding algorithms are proposed according to the idea of significance tree splitting. The algorithms include coding with list using depth-first scanning, coding with list using breadth-first scanning, coding without list using depth-first scanning and coding without list using breadth-first scanning. Among them, the coding algorithm with list has excellent coding performance, and the coding algorithm without list has the advantages of low complexity and low memory requirement.Third, the bitplane organization algorithm in region of interest (ROI) coding is studied. The ROI parameter and the bitplane parameter are introduced to effectively organize bitplanes and by combining the multiple bitplanes interleaving algorithm and the embedded bitplane coding algorithm based on significance tree splitting, the embedded ROI coding algorithm is achieved. Using the coding algorithm, the relation of image region quality and bit rate is effectively harmonized.Four, by combining holographic representation and zerotree preserving partitioning...