The Image Magnification Model Based On The TV Regular And Its Implementation

Resolution is an important factor of the image’s quality. The high resolutionimages has a widespread use in many areas, such as remote sensing, medical, printing,daily photo processing, etc. But with the limitation of imaging conditions, imagingmethod, storage capacity and the network transmission capacity, it is hard to keephigh level resolution. Image magnification, which is also called super-resolutionrestoration, can improve the quality of the image without the need to change thehardware device. So it has the advantages of low cost, fast speed, and can make fulluse of the existed imaging system.In order to improve the image resolution, improve image quality and visualeffects, in this paper, we analyzes the existing algorithms of image magnification,and puts forward several new models to realize image magnification by focuses on theimage processing algorithm based on partial differential equation. What is more,global and local active contour model of the image segmentation is studied. In thispaper, the main achievements showed as follows:1. The analysis and comparison of resize algorithm based on interpolation andwavelet.We have introduced the existed interpolation algorithm and the wavelet algorithm ofamplifying image. We had do experiments to achieve these algorithms, and compare and analysisthese several algorithms. At last, We introduces the latest research dynamic of this area.2. The algorithm of image’s magnification based on filtering modelThis paper introduced the image filtering through diffusion phenomenon, andintroduced the image processing algorithms of isotropic and anisotropic diffusionmodel. Because of the interpolation magnification image has the phenomenon ofsquare slice and edge with sawtooth, we can regard this phenomenon as a kind ofnoise, so we can use the denoising model to filter the results of interpolationamplification. Programming experiment were carried out, and we have obtained theexpected result.3. The image’s magnification algorithm based on the self-snake modelThe self-snake model is a kind of excellent nonlinear filtering model. Highresolution image and image edge detection function is obtained by bilinearinterpolation amplification, which is as the initial estimates of self-snake model to get a iterative solution. In order to get a more accurate solution, the gaussianmoving average constraints condition is be used. This model use the upwind iterativeformat as a solving method, and the programming experiment has a satisfactoryresults.4. Image’s magnification by restoration modelThis paper, we introduces the TV image restoration model-ROF model, and usethe upwind format iteration method to calculate the model. We amplify the image byfill in the need repaired point by use the image restoration method.5. Put forward a amplification model based on image direction angleThrough the research of the image restoration model-LOT model, and inspiredby it, this paper proposes a amplification model based on image direction angle, anduse the augmented Lagrangian method to gain the numerical solution. Theprogramming experiment verify the effectiveness of our model, and the visual effectof enlarge image has obviously improved and the experimental data also show that thesuperiority of the model.6. Improve the model of hybrid segmentation framework of inhomogeneityimageThrough the study of image segmentation, there presents a segmentationframework combined with active contour based on regional and global area tosegment the adaptive image. By bring in the image edge stopping function toautomatic balance the effection of the interactions with local and global grayinformation. The model using fuzzy region competition algorithm to calculate, and dosome segmentation experiment base on some inhomogeneity image, which provedthat the new framework has better segmentation effect.