基于Gabor滤波器和HOG特征的织物疵点检测

针对织物疵点检测问题，提出了一种基于Gabor滤波器和方向梯度直方图（HOG）特征的织物疵点检测算法。首先使用3个尺度、4个方向的Gabor滤波器组对织物图像进行滤波，并做融合处理，增强织物图像疵点区域和背景纹理之间的对比度；然后使用双边滤波减弱图像背景纹理和噪声的影响；最后将图像划分成均匀子块，提取每个子图像块的HOG特征，利用图像疵点区域和背景纹理的HOG特征差异进行阈值分割实现织物疵点的检测。实验选取5种常见织物疵点进行验证，并与传统的Gabor滤波算法进行了实验对比，结果表明该算法可以较好的抑制织物背景纹理的干扰，更加准确的检测出织物疵点。

Fabric defect detection based on Gabor filter and HOG feature

Aiming at the problem of detection of fabric defect, a fabric defect detection method based on Gabor filter and Histogram of Oriented Gradient(HOG) feature is presented. Firstly, a bank of Gabor filters with three scales and four directions are used to process the fabric image, through image fusion, the contrast between the defect zone and the background texture in the fabric image is enhanced; Then, the bilateral filter is used to weaken the influence of the background texture and noise of the fabric image; And Finally, the filtered image is divided into small image blocks with the same size, after extracting the HOG feature of each sub-image block, threshold segmentation is executed based on the difference of the HOG feature between defect area and background texture to realize the detection of fabric defect. Five normal fabric defects are tested, and a contrast experiment with the traditional Gabor algorithm is conducted. The experiment results show that the algorithm can better suppress the interference of the background texture and detect fabric defects more accurately.