A Multi-Level Colour Thresholding Based Segmentation Approach for Improved Identification of the Defective Region in Leather Surfaces

Abstract

Vision systems are recently adopted for defect detection in leather surface to overcome difficulties of labour intensive, time consuming manual inspection process. Suitable image processing techniques needs to be developed for accurate detection of leather defects. Existing research works have focused for gray scale based image processing techniques which requires conversion of colour images using an averaging method and it lacks sensitivity for detecting the leather defects due to the random and texture surface of the leather.  This work presents a colour processing approach for improved identification of leather defects using a multi-level thresholding function. In this work, the colour leather images are processed in ‘Lab’ colour domain for improving the human perception of discriminating the leather defects.  In the present work, the specific range of values for the colour attributes of different leather defect in colour leather samples are identified using the colour histogram.  MATLAB software routine is developed for identifying defects in specific ranges of colour attributes and the results are presented.  From the results, it is found that proposed provides a simpler approach for identifying the defective regions based on the colour attributes of the surface with improved human perception. The proposed methodology can be implemented in graphical processing units for efficiently detecting several types of defects using specific thresholds for the automated real-time inspection of leather defects.