Robust container code recognition system

Manually identification, typically employed when containers enter and exist the port, is inefficient, insufficient and may cause a threat to security. An image-based automatic container code recognition system is introduced that is robust to varying light and shadows. Filtered with both gray gradient information and the characteristics of container codes, the code lines can be properly located. The feature-based Local Intensity Gradient (LIG) and adaptive multi-threshold methods are combined to segment the codes from background with high performance. Using the features of both the contour profiles and points, some damaged codes can be recognized and the correct rate is improved. This system is simple and valid, and gets satisfying result in experiments. Finally, it gives some amendment suggestion.