Automatic Recognition Method of Aviation Thin Cable Characters Based on Rotating Monocular Camera

Abstract

In order to solve the difficulty of manual recognizing the characters printed on thin aviation cables, an automatic recognition method by rotating a monocular camera is presented., two indexes that reflect the completeness and centralizer of characters are designed to automatically search an appropriate image of aviation cable captured by the rotated camera. Then, an optimal image-stitching method is proposed by finding the peak point of ‘coincidence of black pixels’, which improve the quality of character image. Moreover, based on the equal-spaced and straight-line distribution of cable characters, the projection algorithm is optimized, and a character extraction algorithm considering the black pixel’s density and degree of centering is developed. Finally, a-multi SVM classifier is designed to achieve highly accurate recognition of confusing characters. The experimental results demonstrate the effectiveness of the recognition method and algorithm.