Research On Establishing Method And Difference Of Remote Sensing Image Information Capacity Model

The information capacity is a concept of the computer graphics field, It can be used for image quality evaluation by statistics of the cumulative frequency of the local scatter of two-dimensional histogram. The core of the capacity theory is the choice of the Constraint region. Studies have shown that there are two different constraint region selection methods of information capacity. One of them is for image quality evaluation, and another is for the complex evaluation of the image space. The image information capacity of complexity evaluation is an uncertainty measurement of the digital image pixel value distributions. It can reflect the space complexity of surface distribution by courting the discrete degree of adjacent pixel values.In this paper, based on the in-depth analysis of scientific connotation of information cap-acity, a deep study on information capacity model of image complexity evaluation is carried out. Through the combination of GIS window theory, it builds and validates the window infor-mation capacity model. Finally, the information capacity theory is applied to remote sensing analysis to study multi-factor difference of remote sensing information capacity. The results obtained by this paper are as follows:(1) By elaborating the scientific characterization of information capacity theory, two constraint region based on the quality evaluation and the space complexity of the image were proposed. After experimental research for the SPOT5image, we set the parameters of the constraint region. Finally, the algorithm of the constraint region is optimized. The experiments show that the time performance improvement can be made exponentially.(2) Based on the study of window analysis and sliding window, a method for determining the scale of the window is proposed and the window operator is completed, then it is used to sample regions authentication. The experimental results show that the more surface features types of the sample region is, the more distribution complexity is, the more fragmentation degree of surface morphology is, the mean value of window information capacity is greater; the mean value of window information capacity is a changing value, with the continuous change of window size, it increases linearly; The window information capacity histogram reflects the grade distribution characteristics of brightness level complexity of the sample district coverings.(3) Research on difference of remote sensing image information capacity model. It includes different types of surface features, bands, landforms and vegetation cover. Experiments show that:towns, villages and other artificial surface features have high capacity value while the values of farmland, woodland, bare land, waters and others are smaller; The texture features of SPOT5panchromatic image is significant, and it shows high-definition, maximum information capacity value. The information capacity value of green, red and near infrared band are little different, the overall gray value of short-wave infrared band is not high, the Brightness discrimination between surface features is the worst, so the information capacity value is the lowest; Through irregular distribution of vegetation and landforms on the surface, it resulted in the fragmented shadow and complex surface landscape and it enriched gray complexity of the image, thus affecting the information capacity value.