Research And Application Of The Panoramic Annular Lens Imaging System

Compared with traditional optical lens, panoramic annular lens is a novel type. The optical system which is made up of it can observe omni-directional objects around optical lens without any moving parts, so it is very suitable for observing and surveying the objects at the inner surface of a cylinder, such as a pipe. This thesis mainly studies the application of panoramic annular lens on endoscopy of pipes.The thesis is comprised of four parts. Firstly, traditional imaging principle and panoramic annular lens imaging principle are introduced and compared in detail, and then this paper introduces the whole PAL-imaging system and the system's hardware components, then the panoramic images and video are collected from the system. Secondly, the panoramic images are pre-processed, image's gray level transformation and median filter are accomplished and ready for further processed according to the relationship between three color models and the principle of image enhancement. Thirdly, annular images this system acquired have aberration and distortion due to the unique imaging characteristic, thus they are not convenient for observation and surveying. On the base of previous researches on panoramic annular lens, the annular images need to be unwrapped to the rectangular images without distortion according to the annular imaging principle, the problem of the decreased resolution from outer circles to inner ones needs to be solved during unwrapping procedure, this paper adopts a novel interpolation method based on edge detection that adopts cubic convolution interpolation for edge and bilinear interpolation for none-edge. This method can restore the details of images better and decrease the computational cost. When software system of pipes is constructed, a table of trigonometric function is designed to optimize the unwrapping procedures. Practices prove that looking up this table is very efficient in the reduction of calculation and improving this system's performance. Finally, in order to obtain broader vertical vision on inner surface of a pipe, on the basis of the similarity between two adjacent images'gray level, this paper get the overlapped areas between them, and fuse them by weighting factor.