| In this thesis project, we have developed and implemented a prototype for evaluation of natural gas plastic Pipes. We implemented this on raspberry pi for experimental analysis using real time processing. The whole platform consists of 3 inch diameter plastic module built using 3d printer, which holds structured light laser source, a fish eye camera and embedded board for processing and recognizing defects of the inner pipeline structure that is not easily accessible for human inspectors. The embedded machine acquires the data from connected camera via serial bus and follows the 3d rendering algorithm. The algorithm is designed to process the data in real time and render all 2d frames into 3D pipe with exactly same features of pipe. The prototype was developed on a level-by-level approach. Firstly, we implemented using an optical camera with 90-degree field of view that can only cover a part of structured light. Secondly, we use a fisheye camera with 180-degree field of view, which gives desired results of complete ring, which is used as the input of developed rendering algorithm. According to above-mentioned method the motion of camera within pipe is assumed to be constant (10 frames per cm). Finally, the location of cracks within pipes can be detected and the size of cracks can also be recognized using number of frames. The profile of cracks can easily be seen on rendered 3d structure. The raspberry pi is limiting the use of prototype in long length pipes because as the length increase it leads to huge volume of data and can't be handled using limited computational power of pi. The possibility of the future work would include the use of graphic processing unit for faster processing and rendering. |
PDF Full Text Request