Research On Multi-scale Error Diffusion Halftone Algorithm Considering Image Edge Information

Image is an indispensable information carrier in human daily life,work and learning.According to the statistics,about 75%of the information obtained by human comes from vision,and it can be seen that images play an important role in information transmission.There are three processes in the propagation of image information which are acquisition,processing and output.Of the three processes,the output process of the image is an important part of information reproduction.The quality of output process directly determines the accuracy of the image information transmission.Since common image output devices such as printers,printers,etc.are usually binary devices,image reproduction can only be achieved by having two states with ink dots and no ink dots.But the tone of image obtained by image acquisition devices is rich which the output device cannot output directly.So it is necessary to use a digital image halftone method to convert the acquired image into a halftone image.The commonly used image digital halftone methods can be divided into neighborhood process method,point process method and iterative process method.Among these methods,the neighborhood process image halftone method represented by error diffusion has become one of the most popular image halftone methods with its excellent output quality and relatively small complexity.In this paper,to solve the problems of the error leakage in the process of error diffusion which leads to tone darkening of halftone image and ignoring the edge and structure information of the image to determine the order of the halftone processing which causes the edge of the halftone image to remain poor of multi-scale error diffusion halftone method,the following two improvements are proposed:Firstly to the problem of error leakage of the multi-scale error diffusion image halftone method due to not considering the presence or absence of processed pixels in the error diffusion neighborhood of the current processing point.A dynamic non-causal error diffusion filter is designed.The filter designed expands the error diffusion range to improve the visual homogeneity of the halftone image.And on the other hand,dynamically updating the coefficient of the non-causal error diffusion filter can prevent error leakage so as to maintain the tone of the input image;Secondly to solve the problem of poor preservation of image edge and structure information caused by ignoring edge information when halftoing the input image.A new order of halftone determined by combining the gray information and the edge information of the image.The pixel with maximum weighted sum of the edge information and the grayscale information is processed with high priority so as to maintain the edge structure of the original input image.The experimental results show that the multi-scale error diffusion image halftone method considering edge information proposed in this paper inhibits the problem of directional texture and pseudo-contour to some extent which are easily generated in the error diffusion method and guarantees the gray level consistent with the input image.The visual effect of the halftone images is uniform,and the edge consistency of the image is effectively maintained by taking the edge information of the input image into consideration.The edge and structure information in the halftone image is maintained well.