為解決服裝行業實時布匹質量檢測問題，本文提出了一種最優Gabor 濾波器的設計方法。基於布料瑕疵的角度及紋理等獨特性質的分析，該方法利用DE 算法優化Gabor 濾波器参数，從而獲得最佳的瑕疵特征提取。該方法利用两個最優Gabor 濾波器代替一個覆蓋整個頻域的濾波器組，在保證較高的瑕疵檢測準確率的同時降低了系統的計算複雜度。基於真實布料樣本的離線實驗驗證了所提出的最優Gabor 濾波器能有效地應用於瑕疵檢測。
This paper presents a novel design method for optimal Gabor filters for solving real time fabric inspection problem in fashion industry. In this paper, based on analysis of the features of fabric defects, the defect detection model utilizes DE to optimize the parameters of Gabor filters, which helps achieve the best feature extraction of defects. Using optimal Gabor filters instead of a Gabor filter bank, this method is able to obtain high efficiency, and reduce the total computational cost. The performance of the proposed method is evaluated through extensive experiments on real fabric samples.
