应用目标稀少特征的织物疵点图像分割

Image segmentation of fabric defect based on object rarity feature

应用自适应阈值分割方法进行织物疵点检测,其阈值选择一直是研究的难点问题,为此,在研究目标特征的基础上,提出了一种利用目标稀少特征确定最佳阈值的织物疵点图像分割方法。首先,对采集织物图像进行滑动均值滤波,以抑制正常织物纹理信息对疵点检测的影响;然后,对织物图像进行预分割以确定最佳稀少度,在此基础上计算出最佳分割阈值;最后,利用最佳阈值进行织物疵点图像分割,并采用形态学滤波消除伪目标及噪声。结果表明,本文方法能够解决织物疵点分割中的最佳阈值确定问题,能够从织物纹理中有效分割出疵点信息。

Segmentation method based on adaptive threshold is one of the most common detection methods of fabric defects. However,threshold valve selection is a problem in research. Based on the research on target feature,the segmentation method of fabric defect image is proposed by using object rarity feature to determine the optimal threshold in this paper. First of all,the normal fabric texture information is suppressed by moving average filter of acquisition images in order to eliminate the influence of defect detection. Secondly,the optimal rarity is determined by the pre-segmentation and calculated the optimal threshold about fabric image. Finally,the fabric defect is segmented from fabric image using the optimal threshold,and false targets and noises are eliminated by morphological filtering. The test results show that the recommended method can effectively solve the optimal threshold for fabric defect segmentation problem and can segment the defect information from fabric texture effectively.