A Computer Vision Approach for Automatic Dimensional Analysis on Cut Textile Piece

The textile industry is facing increasing quality requirements for its products. This work proposes a computer vision method for Automatic Dimensional Analysis (ADA) on cut textile piece. The proposal includes the combination of several techniques such as homography, alignment, edge detection, to find unknowns defects in the contours of the piece. It is flexible and can be used with great variety of cameras. Experiments showed that the implemented algorithms are able to respond satisfactorily to different shapes and sizes of pieces. Further experiments were carried out to demonstrate the robustness of the proposed method by varying various parameters at different stages of the process.