Remote Sensing Image Enhancement Based On Nonsubsampled Contourlet Transform

On the basis of the peculiarity of remote sensing image and the differences of theexternal environment in imaging, lead to the acquired remote sensing image haveshortcomings of low contrast, edge and detail information is lost and the noise isamplified. As an image processing technique, image enhancement is an indispensablesection. It has many applications range, such as astronomy, agriculture, resourcesurveys, urban planning and industrial areas, Image enhancement brought us greateconomic benefits.In the first chapter of the paper, the growth and research status of imageenhancement in recent years both domestic and foreign were presented. Theclassification of image enhancement methods and some basic image enhancementmethod were described in the second chapter, which mainly introduce the grayscaleconversion and histogram processing technology, In the third chapter of the paper, wepresented Fourier transform, Wavelet transform, Contourlet transform andNonsubsampled Contourlet transform(NSCT), and theirs have features, theadvantages and disadvantages of the algorithm implementation are described andanalyzed. The fourth chapter describes the image sharpening techniques and theunsharp masking algorithm. Finally, Image has low contrast and edge and detailinformation loss phenomenon to solve the problem of the enhanced image has lowcontrast and detail information loss phenomenon, a new method of imageenhancement based on NSCT and unsharp masking was proposed in the fifth chaoter.In this method, firstly, the original image was processed by NSCT transform, andclassified NSCT coefficients into three categories: strong edge coefficients, weakedge coefficients, and noise coefficients. Then, to the strong edge coefficients, thelinear diversification in view of the minimum and maximum of modulus is adopted toenhance the contrast of the original image. To the weak edge coefficients, the unsharpmasking algorithm is adopted to enhance the detail and outline of the original image. We assign zero to the noisy coefficients. Lastly, the NSCT inverse transformation ismade to obtain the enhanced images.To evaluate the enhanced result of the proposed algorithm, definition and meantogether with standard deviation were listed as objective indicators, and visualobservation is also considered. The results illustrate that the presented algorithmobtain better results in enhancing the edge details and the whole contrast of theimages, which also has a better visual effect.