Improved active contour modelling for isolating different hues in infrared thermograms

Abstract

Thermograms are widely used in industries and medicine for quality assessment and diagnostics. Thermogram analysis involves isolating desirable Regions of Interest (RoIs) and characterizing them in terms of physical parameters. Conventionally, image segmentation techniques are developed to extract a specific Region of Interest. However, an efficient algorithm that could segment any desired Region of Interest is yet to be developed. In this study, an efficient color image segmentation technique for isolating RoIs (in infrared thermograms) is developed by using Improved Active Contour Modeling (I-ACM). Two different techniques namely thresholding and region growing, have been used for mask selection in order to increase the segmentation ability of the algorithm. The performance of the proposed techniques is measured in terms of error and computational complexity. A user friendly GUI has been also developed for computer-aided analysis.