Research On Digital Halftoning Technique Based On Error Diffusion

Digital halftoning technique, which is on basis of human visual attribution and color characteristic of image, is the tone rendition technology of transforming an image with greater amplitude resolution to one with lesser amplitude on two levels or on finite levels by mathematics and computer. The technology has widely applied to typography, textile, printing technology, compression memory of digital image, transmission of digital image and medicine etc. The research of digital halftoning technique is important in realistic significance and application value.The thesis studies a series of digital halftoning technique on discussing object of gray image, based on analyzing domestic and foreign digital halftoning techniques. Main contents are as follows: analyzing structure of image rendering system-, discussing on three factors affecting digital halftoning technique which are threshold choice, choice of image scanning path and error filter coefficient; probing into merits and deficiencies of existing technique. In the thesis, many new methods or techniques are:1. Owing to processing result relying on choice of image scanning path, a new halftoning method of context-based space filling curves is proposed. In the method that reflects autocorrelation of image, the optimal path of scanning image is acquired by inherent coherence and relationship among near pixels of image. The smooth property of edge and non-edge region in image is improved, as well as, lagging appearance and regular pattern and structured texture concerning processing direction are reduced.2. In existing extremely main methods, some useful characteristic informationand detail information is deficient rendering, because whole rendering and edge rendering of image are realized by local rendering. Therefore, another new digital halftoning method based on characteristic information of image has been developed. In the method, rendering level to correlative characteristic is enormously raised with three parameters, which are image eigenfunction originated, style of drawing primitives selected, and quantity of drawing primitives confirmed.At last, a great deal experimental analysis and experimental result comparison among above two new methods, order dithering, error diffusion, dot diffusion etc are realized. These experiments had improved that two new methods are superior to existing digital halftoning methods in aspect of subjective sense and objective evaluation, that different parameters are decided according to user requirement and different requisite information of original image are rendered in the new digital halftoning method based on characteristic information of image.