A Simple Gamut Mapping Method Based on BP Neural Network
Very complicated relation lies in gamut mapping between kinds of different prepress devices, which can’t be described properly by linear models. In recent years, neural network theory is applied to describe this relation. In this paper, 1000 color blocks is produced by combination of uniform interval digital image pixel value of each channel in a 24-bit true color RGB BMP format image file, and the image is used to present color in both source device and destination device, i.e. PC monitor and color printer. The source device gamut is divided into microcosmic gamut sections by chroma prediction model via combination of uniform but less interval digital image pixel value of RGB channel, compared with the chroma value in each section and the chroma value produced by color printer via the same 1000 color blocks, the sections with least color error are produced, corresponded and recorded. The recorded chroma value and the chroma value produced by 1000 color blocks on monitor is used as input data and output data to a BP neural network respectively, thus a gamut mapping model is advanced, which is used to map monitor gamut to color printer gamut. The desired printing chroma value can be produced by the corresponding image pixel with printer calibration model. With the gamut mapping model, the experiment shows that the average color error between the two devices is dramatically reduced.
BP neural network gamut mapping micro space sections
Zhao Lei Tang Baoling
School of Printing Engineering, Hangzhou Dianzi University,Hangzhou 310018, Zhejiang Province, P.R.C National key laboratory of pulp and papermaking engineering, South China University of Technology,
国际会议
The 31st International Congress on Imaging Science(第31届国际影像科学大会 ICIS2010)
北京
英文
658-661
2010-05-12（万方平台首次上网日期，不代表论文的发表时间）