Automatic mask creation of thermographic gait images using CVIPtools ATAT

摘要

Thermographic imaging has been shown an efficacious method for analysis of gait abnormality in canines. Gait abnormality in canines results in painful walking, loss of balance, limping, and a number of other uncomfortable and dangerous symptoms. An algorithm was developed for the automatic mask creation of thermographic canine leg images in order to locate important gait related areas. The CVIPtools Algorithm Test and Analysis Tool (ATAT) allows users to examine multiple algorithm sequences, different parameter mixtures, and large image sets within the confines of a single test. ATAT is developed in the Computer Vision and Image Processing Laboratory at Southern Illinois University Edwardsville. Through years of development, ATAT has incorporated 80 different imaging functions from the CVIPtools C libraries. One major testing ground for ATAT is the automatic creation of masks for thermographic veterinary images. ATAT was used to generate potential mask images from multiple enhancement and segmentation algorithm combinations. Error metrics include the Dice coefficient and Jaccard index to compare the resulting images to the ideal hand drawn masks. Using these error results, an algorithm is proposed for the automatic mask creation of veterinary gait thermographic images. A total of 168 images were used from the anterior, lateral, and posterior views of canine legs. The ATAT software was able to identify an algorithm with an average 88.4% success rate between its two-error metrics.