Image Fusion Based On?DCT?Discrete Cosine Transform

The development of remote sensing technology provides an effective technical means for human beings to define different environmental objects and utilize the involving data to extract the required information.There are many types of remote sensing sensors,and different sensors have different characteristics for imaging the same scene,producing such as multispectral images,panchromatic Multi-source remote sensing images,infrared images,SAR images,etc.The image data acquired by various imaging sensors every day is massive,in order to process and use these data efficiently and comprehensively,multiple sources remote sensing images of the same scene need to be registered and fused.Multi-source Remote sensing image fusion uses specific techniques to remove redundant information from images from different sources for the same scene and combines Complementary information to generate a more precise,more accurate,and more comprehensive picture of the scene.Remote sensing image fusion is widely used in both military and civilian applications.First,this study investigated the possibility of enhancing image compression result based on DCT process,where we transformed the image from RGB to YCb Cr,which has a great advantage over RGB,that Cr and Cb components can be depicted with a resolution less than Y because the human visual system is less sensitive to color than to luminance,thus,the subsampling of chrominance type 4: 2: 2 then 4: 2: 0 was applied whereas the image is cut into blocks of 8x8 pixels to apply the DCT to highlight areas with high spatial redundancy.Thanks to the lossy compression,we have been able to propose a quantization mask that eliminates the non-zero high energy values that appear in different parts of the block,to gain the maximum possible zeros adjacent after the zigzag scan.,to get a smaller size using RLE.Also,the DCs coefficients can then be a much smaller weight after the DPCM because they are often close.The Huffman code eliminates the reappearance of values that are obtained from DPCM or RLE.On the other hand,a second operation has the inverse stages called a decoder.We then contrasted the execution time of the aforementioned process with other studies' execution times.The outcome of the analysis shows a marked increase in execution time.In this context,the proposed zigzag table will explain this change as a replacement for loops,as well as the requirements for speed of computation and simplicity of code.All of this reflects positively on overall cost of storage.Second,it applies the DCT technique to the fusion of panchromatic and multispectral images in the remote sensing image.First,perform a DCT transformation,respectively,on the panchromatic image and Y component,which is obtained by performing a YIQ transformation on the multispectral image.Pursuant to the panchromatic image distribution property of the DCT coefficient,classify the DCT coefficient into several classes and divide the DCT coefficient into the low frequency component and the high frequency component by calculating the number of the different class.And then replace the multispectral image's low-frequency component with one of the panchromatic images,and preserve the panchromatic image's high-frequency component.Finally,perform an inverse transformation of the DCT,and the fused image is gained.Experiments have shown that the fused image can successfully reach a trade-off between spectral and spatial information,and that the method is feasible and effective.The proposed enhancement of DCT combined with image fusion,has then implemented through “MATLAB” environment to conduct some experiments over a set of images.Afterward,an optimization process on the code to enhance its functionality has been conducted to make it applicable for use in real-time(compression and decompression).