A method for fast measuring line-shaped image feature based on normalized cross correlation operator

Line-shaped image features are widely applied in many cases and the measurement is required as fast as possible. However, a traditional method (normalized cross correlation operator, NCCO) which is used for pattern matching in microscopic vision needs lots of calculations for a considerable time. Inspired by this, some research work about pattern matching of geometric image features is presented. Firstly, a probability distribution of image features is worked out to describe functions with the operator NCCO through matching two binary images. Secondly, the formulas with linear image parameters are derived and certified by designing special templates and line-shaped features. Based on this model, a measuring algorithm of line-shaped image features is proposed. The simulation studies of the image mixed with moderate-intensity noise conclude that the algorithm accuracy of angle α is up to 0.24°~0.98°. Finally, the algorithm is applied in a hybrid microassembly work-bench to measure the feature's position. Experimental results show that the algorithm based on NCCO mathematic model is more precise than traditional method in real-time process.