Integral images-based approach for fabric defect detection

Abstract

Normalized cross correlation (NCC) technique is the most accurate and robust in defect detection techniques in industry. But its robustness is sensitive against change lighting and the computational cost is very expensive. This paper presents a new approach to the fabric defect detection problem using integral images scheme. The integral image technique is used to substitute the huge operations in NCC to speed up the calculation process. So, the defect detection process can be done in real time. The integration process improves the robustness against different types of noises. The performance of the proposed approach is evaluated by using three different datasets of fabric images with different types of fabric defects. Experimental results demonstrate the proposed approach is efficient and effective for detecting defects in flawed fabric in real time. As a result, obtaining correlation values between windows through integral image transform has reduced computation cost by 98.78% and has successfully provided the detection rate of the defects by 97.5% at minimum.