特征点稀少的炮管内膛图像拼接技术研究

针对火炮身管内膛图像特点,提出一种特征点稀少的火炮身管内膛图像拼接算法。通过Harris角点提取,进行特征点匹配,消除不良匹配点和误匹配点,利用Levenberg-Marquardt算法进行参数优化设计和双线性插值完成图像的重注册,最后采用函数加权法来处理重叠区域的图像,确保内膛图像的无缝过渡。利用Matlab和Visual C 6.0混合编程技术,开发了火炮身管内膛图像拼接软件系统,对内膛检测硬件系统采集到的跨视场内膛图像进行拼接与展开,结果表明,该算法速度较快、稳健性较好、图像配准精度较高。

Image Mosaic Technology for Inner Surface of Artillery Pipe with Small Number of Feature Points

This paper takes advantage of the features of the inner surface image points in artillery pipe,and proposes an algorithm based on small number of feature points on inner surface of artillery pipe.It picks up Harris corner in the first step and matches the feature points later to remove bad and wrong match points.The Levenberg-Marquardt algorithm is used to get the exact transformation matrix M and Bilinear Interpolation technology is used to finish image registration.Finally,an image blending algorithm based on weight means is used to insure non-gap transition in the inner surface mosaicing image.A mix-programming technology is used to develop a mosaicing software system for inner surface image of artillery pipe,to mosaic and expand the span-field view of inner surface images captured by testing hardware system.The experiment results and data show that the image mosaics algorithm for inner surface of artillery can achieve ideal effects,in a short time,and with robustness and matching precision.