Seals in Chinese ancient paintings are rich in semantic information,such as names of authors or collectors. A multi⁃feature based seal partition algorithm for Chinese ancient paintings is presented. The input ancient color painting image is enhanced by correcting background color cast first,and then the enhanced red image is calculated and potential seal regions are detected. Finally,four shape features are employed for verifying the locations of real seals. The experimental results show the proposed method is efficient for locating most square and round seals.%中国国画中的印章含有丰富的语义信息,如画作作者或收藏者的姓名信息。提出一种基于多特征的中国古画印章分割方法。首先,对输入的古画彩色图像进行双边滤波和颜色增强处理,纠正古画的偏色情况,并增强图像颜色的饱和度;然后计算红色通道增强图像,检测印章可能出现的区域;再结合4种形状特征,准确定位印章区域。实验结果表明使用所提算法能够有效定位方形和圆形印章。
展开▼