激光与光电子学进展, 2010, 47 (6): 061001, 网络出版: 2010-05-06
基于不变性联想神经网络的道路交通标志识别算法
Research on Road Traffic Sign Identification Algorithm Based on Invariance Association Neural Network
图像处理 交通标志 模式识别 神经网络 颜色形状模型 image processing traffic sign pattern recognition neural network color-shape model
摘要
提出一种基于不变性联想神经网络的道路交通标志识别算法。该算法通过对交通标志的颜色属性和形状属性的分析,建立了交通标志的颜色和形状之间确定的关系,以其作为识别交通标志的重要依据。建立颜色形状特征库,设计联想记忆神经网络模型用来实现不变性模式识别,采取约束突触权值的方法从图像中提取不变特征。仿真表明该方法可以对交通标志实现快速分类,较好地消除了视角的影响,具有良好的准确性、实时性和稳健性。
Abstract
Feature library of colors and shapes is established. The association memory neural network model is also designed to realize the identification of invariance pattern. The method of constraint synapse weight is adopted to extract invariance feature from the images. The simulation results show that this method can realize quick splitting of traffic signs and the influence of the angle of view has been well eliminated. It has good accuracy,real-timing and robustness.
侯培国, 陈毅强, 张北. 基于不变性联想神经网络的道路交通标志识别算法[J]. 激光与光电子学进展, 2010, 47(6): 061001. Hou Peiguo, Chen Yiqiang, Zhang Bei. Research on Road Traffic Sign Identification Algorithm Based on Invariance Association Neural Network[J]. Laser & Optoelectronics Progress, 2010, 47(6): 061001.