【作者】 吴莹； 梁炯； 殷嘉林； 聂君峪；

【Author】 Ying Wu;Jiong Liang;Jialin Yin;Junyu Nie;School of Printing and Packaging Engineering,Beijing Institute of Graphic Communication;

【机构】 北京印刷学院印刷与包装工程学院；

【摘要】 本文研究通过区域生长算法提取图像金色区域,用于生成印刷使用的金墨色版,使印刷图像更接近真实的金属质感。首先将图像从RGB空间转换到HSV空间,选择H分量作为被分割的原始图像。其次,HSV的H分量存在背景干扰的问题,而区域生长的方法不存在背景干扰的问题,所以可以通过区域生长法分割H分量。最后,结合HSV的H分量和RGB的混合分量得到图像的金色区域。实验结果表明,通过0tsu算法设置阈值比手动设置阈值选定的区域更适合分割金色区域,通过多点区域生长得到的图像比单点区域生长得到的图像更加完整。因此,我们建议采用以0tsu算法设置阈值的多点区域生长法,提取图像的金色区域。从结果看,多点区域生长法分割效果良好,从而验证了我们提出的方法的可行性。更多还原

【Abstract】 The paper proposed an algorithm to extract the con-tone golden region in an image as a printing plate of metallic ink.Firstly the image is converted into HSV color model.H component are selected as the original for image segmentation.Secondly,background interference is a problem for H component of HSV model.Region growing method not interfered by it.We suggest to segment H component by region growing method.Lastly,H component of HSV model and RGB-mixed component of RGB model are combined to get a golden area image.The experimental results show that,it is suitable to determine a threshold value by Otsu algorithm than setting threshold value manually.Multi-point region growing image is more complete than single-point region growing image.Therefore we suggest using the multi-point region growing to extract the golden area of image.It brings good result,which verifies the feasibility of our suggested method.更多还原