Research Of Error Diffusion Algorithm Based On HVS Model

The digital halftone technology is a kind of image reproduction technology. With the help ofauxiliary tools such as mathematics, computer, the technology processes continuous tone image, thenoutputs the representation of the image on the binary devices or multicolored binary devices. In the process,it uses human eyes visual characteristics and color characteristic of image as well. The technology is thecore of the steps in the process of printing, and widely used in image transmission, textile, medical andother fields. So the digital halftone technology has important significance.The main content of this thesis include: introduction to the overview and basic theory of currentdigital halftone technique; the contrast of several classic halftone algorithm. Image processing system is acomplex system, and digital halftone technique is mainly applied to image output section, so this thesisanalyzed image processing system model. Human eyes are equivalent to low pass filter and directlyevaluate the output images, so this thesis introduced human visual properties and human visual system(HVS). In addition, this thesis did highlight introduction and in-depth research on the basic idea and threemain influencing factors of error diffusion algorithm.Firstly this thesis has carried on the analysis and research on advantages and disadvantages of thestandard error diffusion digital halftone algorithms. The Halftone image that was processed by standarderror diffusion algorithm has rich display color, but it has more "turtle grain" as well. In the highlights andshadows standard error diffusion algorithm appears lag phenomenon of halftone processing direction; andthe edge part is easy to produce the discontinuous transition phenomenon. According to the aboveshortcomings, this thesis proposed an improved idea: firstly, it did edge detection for the image, in differentareas of the image using different error diffusion algorithm to process, standard error diffusion algorithmprocessed the pixels of un-fringe region and use the error diffusion method with error diffusion coefficientwhich based on HVS model. The algorithm keeps the advantages of the standard error diffusion algorithm,at the same time, reduces the false contour phenomenon, as much as possible to make the edge informationcomplete. After that it puts HVS model action on original continuous tone image and optimized image by error diffusion algorithm, and gets the new image and judgment of similarity. On this basis it rectifies theimage processed by error diffusion.Finally, this thesis summarized several common and existing image quality evaluation methods,introduced several common image quality evaluation methods, and analyzed their advantages anddisadvantages, and used LiveDatabase for experiment. According to the human eye features, it put forwardthe image evaluation method based on HVS.