| Recent years have seen a rapid growth of interest by the scientists of China, America, French, Japan and other countries in inertial confinement fusion(ICF) for ICF is the most hopeful method to control nuclear fusion and use the energy. The key point of ICF is the detection and matching of ICF targets. Now in China the usual method is measuring offline and matching online with some disadvantages like cumbersome steps and long time-consuming.In this thesis, a 3D-position-detecting method based on CAD model and machine vision was presented to orientate some targets with complex shape. By this way, the position detecting can be realized online and get attitude information of the target by matching CAD models with photographs taken by CCD. To solve the problem that short depth of field caused defocus, three detail solutions were put forward to restoral the fuzzy graphs. The first step is to fuzz photographs artificially and compare these photos with those got previously. Then select the most similar one and restoral the fuzzy photos. The second step is using BIR(Blind image restoration) algorithms to iterate and update the high-frequency parts of fuzzy kernels and photos and then restoral photos. The last step is to gather the square waves in different locations by CCD and the computer simulation of the point spread function and optical transfer function are done in separate parts. Spliced images and then the whole target photographs will be obtained.At last, the photographs obtained by different methods were analyzed by posture matching experiments, in which the matching results were compared with each other. And the detail parameters were analyzed and then the final plan which could guarantee matching speed and accuracy was obtained. |
PDF Full Text Request