High-definition 3D Electronic Endoscope Image Processing System Research And Development

With the continuous development of modern science and technology, and mutual penetration between disciplines, high-definition electronic endoscope has become one of indispensable instrument in medical examination and treatment. High-difinition 3D electronic endoscope which can provide three-dimensional visual images has been becoming the current hot spot. Through electronic endoscope for HD 3D imaging, clear images, to provide 3D display, and reconstruction of 3D depth information of the target zone, can effectively improve the accuracy of the medical diagnosis and treatment process. This paper mainly revolves around the hd 3D electronic endoscope image processing system set up, auto white balance algorithms and stereo matching algorithm calculation depth these three aspects.This paper based on the current status of the development of medical endoscope and use requirements introduced high-definition 3D electronic endoscope image processing system. Include software function, the pretreatment of the image data and the realization of the dual channel image data synchronous display. The image data preprocessing can realize automatic calibration, automatic metering, etc. Using the zhang principle for camera calibration, complete the polar image correction. Then according to different lighting, automatically adjust the image exposure, avoid the phenomenon of the image the image or image, for subsequent stereo matching to do a good job of premise.Then, due to the different light source color temperature image obtained by endoscopic system deviates from the true color would happen. In order to improve the imaging quality of endoscopy, we have done some research on auto white balance algorithm in this paper, combined with the algorithm for calculating grayscale world algorithm and theory of knowledge, puts forward the automatic white balance algorithm based on colorless difference surface recognition. The sum of the algorithm by RGB three-channel the colorless almost image are extracted, and by adopting the idea of the weights to obtain colorless difference pixel sets, avoids the threshold selection and makes the algorithm more generality. Algorithm according to get no pixel sets to estimate the illumination chromaticity color, and diagonal transformation after correction according to the image color, color as far as possible, and the algorithm is relatively simple, small amount of calculation and can meet the requirements of real time, in order to achieve the effective, real-time automatic white balance effect.Finally, in stereo matching algorithm research, in view of the situation our system requirements and the actual use, combined the existing stereo matching algorithm, this paper proposes a simple and relatively quick stereo matching algorithm, and add the matching error detection and correction module. Through preliminary stereo matching algorithm based on the simple SSD, the parallax error matching detection, after matching error correction to obtain more accurate parallax figure, this method can meet the real-time performance required by the system at the same time, also to a certain extent to ensure the reliability of the matching and depth information reconstruction. Finally by filming system actual scene graph, calculating the depth information, and the testing of the accuracy of the algorithm validation show that the scheme is feasible.