Automated rivet detection in the EOL image for aircraft lap joints inspection

An enhanced visual inspection technique named Edge of LightTM (EOL) was invented and developed at the Institute for Aerospace Research, National Research Council Canada. One application is the detection and measurement of “pillowing” deformation caused by corrosion in the faying surfaces of riveted aircraft fuselage joints. To quantify the hidden corrosion products, one solution is to relate the average corrosion in a region to the peak-to-peak amplitude of the EOL image brightness profile between two diagonally adjacent rivet centers. This raises the requirement to automatically detect and locate the rivets in an EOL image. The variation in illumination over individual rivets degrades the performance of most of the edge detection algorithms. In this study, a phase congruency algorithm is employed to provide an absolute measurement of EOL image features. The detected feature together with the knowledge of rivet size and the number of rows are used to locate rivets in the EOL image. The performance of the approach is compared with commonly used circle detection algorithms and an adaptive threshold approach. The efficiency of the proposed algorithm is assessed experimentally and the results are used in the quantitative evaluation of aircraft hidden corrosion.