基于Otsu和小波算法实现肖像照黑白木刻风格  

Realization of Converting Portrait Photos into White-line Woodcut Style Picture Based on Otsu and Wavelet Algorithm

作　　者：王晓迎[1] LIU Sixia 黄源[3] 

机构地区：[1]西安思源学院,陕西西安710038 [2]法国利摩日大学,法国利摩日870001 [3]西安邮电大学,陕西西安710061

出　　处：《计算机技术与发展》2016年第1期58-64,共7页Computer Technology and Development

基　　金：陕西省教育厅专项科学研究项目(2013JK0480)

摘　　要：为了将现代科技手段融入到传统手工木刻之中,提出一种实现人物肖像照片黑白木刻风格化的算法。首先利用改进二维Otsu法求出"主阈值"并分割照片,其分割效果关联照片特定局部细节和整体纹理。在基于统计原理的最优区间内选择多个"分阈值"分割照片并设定其灰阶颜色,再与"主阈值"分割后的图像融合,形成照片的"木刻纹理"。然后利用改进方向小波边缘检测算法,对沿小波变换方向搜寻到的极大值进行加权,使检测到的图像边缘宽度和细节数量可控。对分阈值融合图和照片原图进行方向小波边缘检测,其结果融合形成照片的"木刻边缘"。最后将"木刻纹理"和"木刻边缘"融合形成照片的黑白木刻化风格。实验结果表明,该算法具有较好的适应性和实用价值。In order to integrate modern technology into traditional manual woodcut,an algorithm specialized in converting portrait photos into white- line woodcut style picture is developed. First,using an improved two- dimensional Otsu method to obtain the＂main threshold＂to segment the photo,the result is related to the photo ＇s details at specific local areas and the overall texture. In the optimal range based on the principle of statistics,＂sub thresholds＂are made in order to segment the photo and set grayscale color,the segmented pictures fuse with the one that segmented by ＂main threshold＂,forming the photo ＂woodcut texture＂. Secondly,an improved directional wavelet edge detection algorithm is used,along the direction of the wavelet transformation,the maximum value is weighted so that the detected edge width and the detail numbers are controllable. The algorithm is applied to the＂sub thresholds＂fused picture and the photo,the results forming the＂woodcut edge＂. Finally,＂woodcut texture＂fuses with＂woodcut edge＂,forming the photo white- line woodcut style picture. It is shown in the experiments that this algorithm is flexible and practical.

关 键 词：二维OTSU法 方向小波 主阈值 分阈值 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]