This paper describes a vision-based fabric inspection system that accomplishes on-loom inspection of the fabric under construction with 100% coverage. The inspection system, which offers a scalable open architecture, can be manufactured at relatively low cost using off-the-shelf components. While synchronized to the motion of the loom, the developed system first acquires very high-quality vibration-free images of the fabric using either front or backlighting. Then, the acquired images are subjected to a novel defect segmentation algorithm, which is based on the concepts of wavelet transform, image fusion, and the correlation dimension. The essence of this segmentation algorithm is the localization of those events (i.e., defects) in the input images that disrupt the global homogeneity of the background texture. The efficacy of this algorithm, as well as the overall inspection system, has been tested thoroughly under realistic conditions. The system was used to acquire and to analyze more than 3700 images of fabrics that were constructed with two different types of yarn. In each case, the performance of the system was evaluated as an operator introduced defects from 26 categories into the weaving process. The overall detection rate of the presented approach was found to be 89% with a localization accuracy of 0.2 in (i.e., the minimum defect size) and a false alarm rate of 2.5%.