Statistic based iterative method to measure inner radius of tube

Abstract

In this paper, we proposed a novel tube-radius measuring algorithm which has a capability to remove added noise effectively as well as measuring the radius of small tubes within an allowable precision. First, in the algorithm we select a ROI including possible true center of the tube and find a candidate true center in the ROI that minimizes standard deviation with respect to the radius of the tube circle. Second, we remove noise data that is far from the average radius and replace the data with a new one which gives an average radius. The disconnection of data points developed in the replacement is cured by a connection algorithm. The final step of the algorithm is repeating the process until the value of standard deviation drops to a small predefined value. To show the measuring performance of the proposed algorithm we carried out various experiments in both circle geometries with added noise and real tube with complex noise that is assembled to braking unit of automobiles. From the experiments, it is concluded that the measuring precision of the algorithm is recorded to within 1.4% of error even in the 15% of added noise.