基于多参数Gabor滤波器融合和高斯差分的生丝疵点分割OACSTPCD
Segmentation of raw silk defects based on multi-parameter Gabor filter fusion and difference of Gaussian
针对生丝疵点图像背景与疵点差别极小的特点,提出一种基于多方向、多频率Gabor滤波器融合和高斯差分滤波的疵点分割算法.根据疵点纹理特性,首先采用极大值的方法对特定频率多方向疵点Gabor滤波图像进行融合,并基于均值法对不同频率Gabor滤波器图像进行融合,然后使用高斯差分滤波进一步凸显疵点,最后采用全局阈值将疵点与背景分割开来.测试结果表明:该研究算法召回率达到了93.2%,能够准确分割出生丝疵点.
In view of the small difference between background and defects of raw silk image,defect segmentation algorithm was proposed based on multi-directional and multi-frequency Gabor filter fusion and difference of Gaussian filter.According to the texture characteristics of defects,first of all,multi-directional defect Gabor filter images with specific frequency was fused based on maximum method.Gabor filter images with different frequencies were fused based on mean method.Then difference of Gaussian filter was used to highlight defects further.Finally,defects were separated from background by global threshold.Experimental results showed that the recall of the proposed algorithm was reached 93.2%and the proposed algorithm can segment defects accurately.
杨言语;李子印;汪小东;叶飞;姚晓娟;金君;曾凡高
中国计量大学,浙江杭州,310018湖州市质量技术监督检测研究院,浙江湖州,313099
轻工业
生丝疵点分割多参数融合Gabor滤波器高斯差分全局阈值分割
raw silk defect segmentationmulti-parameter fusionGabor filterdifference of Gaussianglobal threshold segmentation
《棉纺织技术》 2024 (006)
63-68 / 6
浙江省市场监督管理局青年科技项目(QN2023446);国家市场监督管理总局科技计划项目(2022MK048);浙江省基础公益研究计划项目(LGN20F50001)
评论