Improved alternative for detecting SMC tape in TSLF cables using LBP descriptors

Abstract

During the manufacture of the TSLF cable, in one of the last phases of production, the cable must be covered with a sealing semiconducting tape (SMC Tape), the application of this tape, can cause defects in the manufacturing, currently there is no system of detection for this defect. This thesis aims to study, develop and test a feasible detection system that can improve the current quality control and reduce the cost associated if the defect occur. Before getting the final working algorithm that better fits the needs of this system, research and test was done in order to find a reliable and simple solution, firstly all the tests were developed with MATLAB thanks to its huge package of pre-build functions, once the final algorithm was defined, the first iteration of the code was translated to Python, a more versatile, light and fast programming language. The results show, a good performance over each iteration. The algorithm perform a good overall detection but it is not robust enough to be implemented in the industry sector. This system is a starting point to improve the current system, although a new software revision is under evaluation (June, 2019). The current system, has a in-build database, which saves al the photo frames, the big future step is to train an artificial intelligence to improve even more its performance and increase the varaivility of the detections.