The Generation And Compression Method Research Of Remote Sensing Image Pyramid Data

Over the last decade, with the ever-changing of technology, remote sensingtechnology is evolving constantly. In particular its resolution requirements arehigh, and the higher the resolution, the greater the amount of image data, whichresulted in the accumulation of vast amounts of remote sensing data. How tocompress and display the large amount of remote sensing image data efficientlyis particularly important. One effective method is to use the pyramid model fordata compression.Pyramid is a multi-resolution level model, which wasconstructed using magnification, thereby form a plurality of resolution levels.From the bottom to top, the resolution of level gets lower and lower.Through sampling, re-sampling to obtain different resolutions images, andthen to build the pyramid model. Build pyramids data model includes threeaspects, the first determine the number of levels of the pyramid, and secondsampling, re-sampling methods, the third compression methods. Re-samplingmethods generally include three types: the nearest pixel interpolation, bilinearinterpolation, bi-cubic interpolation. The most commonly used compressionmethod is JPEG compression. However, the sharpness of multi-resolution imagegot by using existing re-sampling methods is not high, the obtained data compression ratio by using JPEG compression is also to be further improved.Based on the arose problems above, the main purpose of this project is todesign a new pyramid data compression method, which improves the resolutionof each image quality while also greatly limits to improve the compression ratioof the pyramid data. The paper carries out research according to codingcompression technology. The main research work is as follows:By study of traditional hierarchical scalable coding architecture deeply,develop a new generation and compression approach of pyramid data model. Onthe one hand is to re-sampling process: use AVS predictive information in theimage, estimate roughly of the direction according to image grayscale texturechanging. Carried out sampling, re-sampling according to the grain direction,maximize the retention of the upper layer of the image information, therebygreatly improving the quality of the layers of the image resolution, so that thesharpness of each recovered pyramid image can be improved. On the other handis compression of data pyramid: traditional pyramid model, the data of eachlevel need to be compressed encode, data redundancy between the layers is large.This new design of the pyramid model only need to encode the level residualsexcept the topmost layer which the entire data needs to encoded. Residuals areobtained by the layer of the original image with its own down-sampled imagedata back up again sampled differencing. the obtained residual data value in thismanner is small, To further improve the data compression ratio. In the end,through bitstream decoding it can restore each layer picture.