改进频率调谐显著算法在疵点辨识中的应用

为提高疵点检测效率和准确率，提出用改进频率调谐显著（FT）算法替代 Gabor 小波方法预处理疵点图像，强化疵点特征向量灵敏度。分析了FT 算法中高斯滤波器模板、Lab 颜色空间、高斯滤波图像中椒盐噪声和 HSV 颜色空间不同通道取值范围不一致对疵点识别的影响，并提出了相应改进方法。利用改进 FT 算法进行图像显著处理；使用灰度共生矩阵方法对疵点显著图进行特征提取；利用概率神经网络分类器分类，检测是否存在疵点。对 2 种不同纹理面料的检测结果表明：改进 FT 算法较改进前计算时间增加约8%，但疵点检测准确率提高18% ～25%；与 Gabor 小波相比，检测准确率基本持平，但计算时间缩短约70%。

Application of algorithm with improved frequency-tuned salient region

In order to improve the efficiency and accuracy of fabric defect detection, a novel defect detection algorithm with improved frequency-tuned salient region (FT)to replace theGabor wavelet method was proposed to improve the contrast ratio of fabric defection image and enbance the sensitivity of feature vectors.The influence factors of FT algorithm on therecognition precision of fabric defect including Gauss filter template, Lab color space, the salt and pepper noise in the image of the filter and the different ranges of HSV color space were analyzed. The FT algorithm wasimproved based on the analytic result.The fabric images were highlighted by the improved FT algorithm. Simultaneously, the gray-level co-occurrence matrixmethod wasto extract the feature vector from the highlighted image. Finally,probabilistic neural networkwas employed to detect the defect on the fabric image. Through the detection of 2 kinds of fabrics with different textures, the experimental results show that the computational time of the improved FT algorithm is prolonged by about 8%. Meanwhile, the accuracy of defect detection increases by 18%~25%. Compared with the Gabormethod, the detection accuracy is substanially the same, but the computation time is shortened by about 70%.