Research Of Inverse Halftoning Based On Adaptive Median Filtering

Digital halftone is the technique that transforms continuous imageinto binary image by quantitative operation, and the inverse halftone isthe inverse halftone process, that is to say the process of transforminghalftone image into continuous image is called inverse halftone images.In real life, we often can only obtain halftone image, if we want to editimages such as classification and recognition, data compression, imageenhancement and so on, the halftone images must first undergo inversehalftone processing.In this paper, inverse halftoning algorithms of grayscale images andinverse halftoning image quality assessment were studied, and the mainresearch work is as follows:1) An inverse halftoning algorithm based on adaptive median filterwas proposed. The algorithm is mainly aimed at improving the loss ofimage caused by the median filter of the detail fast inverse halftonealgorithm. The algorithm makes use of adaptive median filter to editimage instead of median filter, improves the adaptive median filter, takesdifferent treatment strategies for noise point pulse and the pulse points byjudging the relationship of the pixel’s8direction gradient, enhances theability of adaptive median filter to judge and remove noise,and thusimproves the ability to keep the details of the fast inverse halftonealgorithm.2) An evaluation index based on information entropy weighted andgradient structural similarity was put forward. And the index introducestwo concepts-gradient and information entropy into the structuresimilarity theory, and constructs the corresponding gradient structuresimilarity function to replace the original structure similarity function,and the index processes the image by n*n block, and the processdetermines the weight of the pixel block by calculating the informationentropy of each pixel block, and the experimental results show that thismethod can effectively evaluate the reserve details level of the inversehalftone image.