Research On Information Quantity Of Romote Sensing Imagery

Scientific technology is in the stage of momentous transformation. The interpenetration and amalgamation among disciplinaries have propelled the emergence and development of a multitude of interdisciplinaries. The introduction of information theory into mapping and remote sensing to investigate the uncertainty existing in these fields has become an important direction. It has proved that uncertainty of remote sensing objects objectively exists. How to quantitatively evaluate the uncertainty in remote sensing process and design a uniform measure between information amount and uncertainty is the keystone of this paper.Based on the information theory and probability theory, a main line on how to evaluate uncertainty and information quantity of remote sensing imageries has spread out. Through the analysis of characteristics of information sources and factors influencing the measurement of image information, a set of information measurement indexes has been proposed and calculation methods about information quantity and uncertainty have been discussed, which have been applied to lossless compression of remotely sensed imagery and gained favorable results.The main work and innovations are as follows:1. Based on the information theory, this paper systematically analyses the characteristics of information source, channel, and sink of remote sensing imagery, investigates processes which affect the information content and the quantitative methods for calculating the uncertainty degree of remote sensing imagery, and brings forward a relatively complete index scheme for characterizing remotely sensed information.2. From the perspective of information theory, the characteristics of the information source of remote sensing imagery, which is a kind of memorial information resource and accords with the statistical characteristics of Markov information source, are analyzed. Factors which affect the information content include quantization levels, noise, geometric aberration, correlation among signals and so on. The information redundancy caused by pixel correlation and spectral correlation is analyzed and methods which can be used to eliminate the two kinds of correlation during the procedure of calculating information content are proposed. Moreover, a formula for calculating the position information content of remote sensing imagery is derived.3. The source and characteristics of noise in remote sensing imagery are analyzed, with the concept of noise equivocation being further expatiated. Experiments are carried out to validate the formulas for calculating three kinds of noises which affect imagery information content. Factors that result in geometric aberration are analyzed and formula for measuring the displacement equivocation is deepened through the validation of experiments.4. Multisource remotely sensed images are selected for calculating and comparing the information content. Meaningful conclusions are obtained, which validates the scientific nature and rationality of the information measure theory. At the same time, effects on information content caused by imagery processing are discussed. The principles and functions of several commonly used image processing methods are explained from the perspective of information theory.5. The principles of image compression are analyzed and currently used lossless and lossy compression methods are retrospected. The essence of image compression is to eliminate correlation. According to the statistical rule of remote sensing information content, differential coding compression method is proposed. When this lossless compression method applied to the multi-spectral data and panchromatic band data tanking from "Beinjing No.1" small satellites, favorable results are obtained. The proposal of District Forecast differential coding compression is brought forward.