Image Processing Software Research And Development For High-definition Endoscope Based On Multi-core Processor

Medical electronic endoscope is a vital instrument for clinical examining and treating. With the introduction of high-definition television (HDTV) technology to endoscope video processing system, medical electronic endoscope products with1080p definition and30fps frame rate has emerged; however, the special application environment of endoscope calls for better dynamic effect and stability, which leads to further study and development of endoscope video processing system with1080p definition and60fps frame rate and efficient image coding method for key frame compression and storage.This thesis develops image processing software, including image front-end processing software and JPEG XR coding/decoding software for endoscope video processing system with1080p definition and60fps frame rate on TlLE-Gx8036multi-core processor platform.Firstly, on the basis of a detailed analysis of each building block in the image front-end processing software’s, this thesis designs image front-end processing software with function dividing method on8-core parallel architecture; then this thesis optimizes the software with such TILE-Gx8036processor characteristic as assembly, compiler and storage optimization. Experiment results shows that the software is able to process1080p image with a rate of61.8fps and a speedup ratio of4.45.Moreover, on the basis of a detailed analysis of JPEG XR image compression standard characteristic, this thesis designs JPEG XR image coding software with function dividing and data dividing method on30-core parallel architecture; experiment results shows that the software is able to compress1080p image with a rate of more than20fps and a speedup ratio of14.17at most. This thesis also designs JPEG XR image decoding software with function dividing method on6-core parallel architecture; experiment results shows that the software is able to decompress1080p image with a rate of5fps at most and a speedup ratio of3.14at most.Finally, this thesis summarizes the study and gives a prospect of future work on how to improve the image quality for image front-end processing software and improve the speedup ratio for JPEG XR coding/decoding.