Color-strip Recognition Based Information Management System: Implementation on OBD Management

Zhipeng Zhang, Yuan Hao, Xiaohang Yuan, Wenting Ma, Tieguang Cai, Hongyi Tang, Jun Xiao, Yingteng He, Xingang Chai, Wei Liu, Yadan Luo, Zhiheng Zhou

Abstract: AI is being used more and more widely in our daily life and industry. Taking optical branching device (OBD) unit as an example, this paper designs an information managing system based on color strip recognition. OBD could be referred as a multiple-in multiple out (MIMO) system managing the optical fiber cables. It is usually used for optical signal coupling, branching, and distribution and mainly for business broadband management and maintenance. The proposal is based on a designed color strip which composed by several color rings and the rings in different colors stand for binary values and the head-tail signs respectively. Using this method in OBD port recognition so that available port resources could be detected automatically through these binary color rings. Due to different illumination conditions, the color distortion caused by the lens and the shooting angle makes the observed color be different from the designed color. Besides, some color strips might be overlapped or curved under certain circumstances. To tackle these problems, we applied a classification method to choose a suitable color combination. Then, an object detection method is applied to detect the color strip and the binary information could be captured. Finally, we proposed an algorithm utilizes the preliminary knowledge (position continuity, the number, and the size consistency of color rings) to achieve the color strip recognition with high accuracy. Our experiments indicate the efficacy of the proposed method.