Design And Analysis Of Video Laparoscopic Digital Image Processing

With the widely application of laparoscopic surgery, it is necessary to develop a medical digital image processing system of television laparoscopic. The Laparoscopic Image Processing System in this thesis sets video image acquisition, image processing, image measurement, image storage and report printing functions into one, in which image processing plays a important role.This thesis has designed a multi-Video laparoscopic digital image processing System, which is based on the platform of Microsoft Visual C++6.0. Many Classes have been encapsulated in MFC(Microsoft Foundation Class Library), which provides a great convenience. The interface of this software system is friendly and easy-to-use, and it includes many important functions to assist doctor to diagnose, like Zoom, Pseudo-color, True color to gray, Anti-color and Sharpening, Brightness adjustment, Mirror Image processing , and the regional arbitrary images perimeter and area measurements.In addition, this thesis deeply analyses many de-noising methods of laparoscopic images, which includes linear filtering, non-linear filtering, self-adaptable filtering, gauss filtering and some low-pass filtering methods in the domain of frequency. This clinical result shows the PSNR (Peak Signal Noise Ratio) in the method of super-neighborhood average filtering is the biggest. As the wavelet transform's multi-resolution feature, it can distill most useful image from the noising image. The method of combing the wavelet threshold method with super-neighborhood average filtering was put forward. The clinical result demonstrates that this method has effectively restrained the noise and its PSNR is bigger than traditional method. Many wavelet functions have been applied to the image, which includes Haar, Daubechies, Coiflet and Symlets. And the result showed the PSNR increases while the vanishing moments increases, and the coif5 performs the best within all these wavelets.