摘要： 通过分析皮棉图像色度分量的分布规律，提出一种皮棉异性纤维特征提取方法，以Cg分量代替Cb分量组成YCrCg颜色空间进行特征提取。为有效地融合各色度特征，采用保矩算法对色度分量进行降维得到一维融合色度子图像，通过对该子图像进行阈值分割最终得到异性纤维特征。实验结果表明，该方法能有效地增加皮棉与异性纤维的对比度，平滑背景干扰，提高异性纤维特征的提取精度。

关键词: 异性纤维, 特征提取, 保矩算法, 色度

Abstract: This paper presents a feature extraction method of foreign fibers in lint cotton by analyzing chrominance distributions of cotton image. Cg is substituted for Cb and then the YCrCg color space is formed to extract the foreign fiber features. To fuse the chrominance features effectively, the moment-preserving principle is employed to realize the dimensionality reduction. The final results of foreign fiber features are obtained by the threshold segmentation of the fusion image. Experimental result shows that it is useful to enhance the contrast between cotton and foreign fibers and smooth the background interference, which helps to improve the extraction accuracy of foreign fiber features.

Key words: foreign fiber, feature extraction, moment-preserving algorithm, chrominance

中图分类号: 

• TP273