A video capture and processing system for laryngoscope diagnosis

Abstract

An implementation of a video capture and processing system for laryngoscope diagnosis based on DirectShow is put forward in the paper. It adopts WDM (Windows Driver Model) as the video capture method, which can enhance the portability and generality of the video capture. It utilized MPEG to encode video stream, which can result in higher encoding rate and better encoding quality. Based on DirectShow Framework, the video capture and processing system for laryngoscope diagnosis has realized the real-time observation and diagnosis for larynx disease. Also, the system utilizes a new glottal segmentation method based on a local-based active contour framework in automatic glottis detection. Comparisons with the previous similar segmentation method resulted in much better performance, while comparisons with manual segmentation resulted in less than 5% difference on the average glottal area. The reliable results can supply accurate information for doctors and otolaryngologists.