基于BP神经网络的集装箱箱号识别

提出一种用BP神经网络实现集装箱箱号识别的方法。提出的图像预处理方法对管柱,底面凹凸,脏污物、光线的强弱、倾斜、缺损或模糊有较强的适应性。在BP网络分类器实现时,考虑集装箱箱号的组成特点,采用并行处理方法。在BP算法中,引入动态因子,动态调整学习步长。实践证明利用这种方法来识别装箱箱号,速度快,识别率高,具有较高的实用价值。

Container Number Recognition Based on BP Neural Network

A method of recognizing container number based on BP neural network is proposed. The proposed image preproeessing method is suitable for columns, concavity, convexity, smudginess, light intensity, slope, imperfectness and faintness. In the realization of BP network classifier, the characteristics of the container number are taken into account and the parallel recognition method is adopted. The dynamic gene and the dynamic adjustment of learning steps are introduced to the BP algorithm. The proposed recognition method is not only fast but also accurate with a high practical value.