基於光影及梯度變化做瑕疵分析

紡織產業是臺灣的傳統產業，近年來往精緻化、差異化發展，隨著運動風潮興起，機能性紡織的重要性也隨之水漲船高。臺灣紡織產業在機能性紡織上有很好的表現，提供了全球知名運動品牌七成以上機能性布料。絲餅為紡織產業的半成品，其製程大部分都已自動化，唯獨絲餅的瑕疵檢測部分仍倚賴人力進行目視檢測。為了降低人力成本、提高檢測品質，我們希望能以自動化輔助檢測，降低檢測人員工作量並提升其效率。本論文提出兩種方法，分別基於光影變化及梯度來實現外溢瑕疵的檢測。我們分別利用不同閥值的二值化影像及HOG分析，來檢測絲餅的異常區域。實驗結果顯示本文的方法可提供自動化檢測瑕疵一個很好的參考，其中包含影像的拍攝方式以及參數的設置。

關鍵字

光影；梯度；瑕疵；絲餅；檢測

並列摘要

The textile industry is a traditional industry in Taiwan and has developed in recent years along directions of refinement and differentiation. With the rise of sports-related applications, the importance of functional textiles has also increased. Taiwan’s textile industry has been do-ing very well as it provides 70% functional textiles for world’s leading sports brands. Bobbin is the work-in-progress production of textiles, with most of its production process being au-tomated, except the inspection of bobbin defect. In order to reduce the labor cost and improve the quality of inspection, we hope to use automatic detection method to aid inspection so as to reduce the workload of inspection technician while improving their efficiency. In this study, we propose two methods to perform bobbin defect analysis based on shadow and gradient. We use binary images from different thresholds and HOG analysis, respectively, to find the abnormal area of the bobbin. Experimental results show that the proposed approach can in-deed provide a good reference for automated inspection, which includes proper procedures for image collection as well as parameter setting.

並列關鍵字

shadow ； gradient ； inspection ； defect ； bobbin

參考文獻

[1] A. Kumar, “Computer-vision-based fabric defect detection: A survey,” IEEE Trans. Ind. Electron., vol. 55, no. 1, pp. 348–363, Jan. 2008.

Google Scholar

[2] H. Y. T. Ngan, G. K. H. Pang, and N. H. C. Yung, “Automated fabric defect detection-A review,” Image Vision Comput., vol. 29, no. 7, pp. 442–458, 2011.