增强现实飞机维修过程中角点检测方法优化

Optimization of corner detection method in augmented reality aircraft maintenance

针对增强现实飞机维修中角点检测方法存在的问题,提出一种角点检测优化方法。用维纳滤波复原样本图像,通过8邻域比较法滤掉约占总像素点60%的非特征点得到初始点集;建立以角点质量等级Q和相邻角点最小距离限度D min双参数为核心的改进Harris算法,借助Matlab分析得出最优的Q值和D min值,将最优参数带入改进算法后可得到更加准确、分散的角点。实验结果表明,该优化方法与传统方法相比在速度和准确性上有了明显提高,利于后期样本图像与维修现场间的实时匹配。

Aiming at the problems of corner detection method in augmented reality aircraft maintenance,an optimization method of corner detection was proposed.The sample image was restored through Wiener filter.By means of 8-neighborhood comparison method,the initial point set was obtained by filtering out the non-feature points that account for 60%of the total pixels.An improved Harris algorithm based on the two critical parameters,corner quality grade Q and the minimum distance between adjacent corners D min was established.The optimal Q and D min were obtained by means of the analysis of Matlab.After the optimal parameters were brought into the improved algorithm,more accurate and dispersed corners were obtained.Experimental results show that the speed and accuracy of the optimization method are significantly improved compared with the traditional method,which is conducive to real-time matching between sample images and maintenance sites.