基于小波神经网络的路面破损识别

              肖旺新^[1][3] 张 雪^[2] 黄 卫^[3]

                                  ( 嘉应学院计算机系¹  东南大学计算机系² 东南大学智能运输系统研究（ITS）中心³)

摘要： 结合图象处理、模式识别等先进技术开发路面破损自动检测系统已经成为本领域的研究热点^[1~5]。本文主要研究了小波神经网络在路面破损识别中的应用，并与传统的BP神经网络作了对比。试验结果表明，在相同的训练样本情况下，小波神经网络的精度高于BP神经网络。为开发更为高效的路面破损自动检测系统提供新的思路。

关键字：公路破损检测，图象处理，破损识别，小波神经网络

                                              Preliminary Study of Pavement Surface Distress Automation Recognition

                                                Based on Wavelet Neural Network

                                                                 XIAO Wang-xin, ZHANG Xue, HUANG Wei

Abstract:  Automatic pavement surface distress survey system based on image processing and pattern recognition has become the hotspot in its field^[1¬5]. In order to improve the accuracy and efficiency to identify the asphalt pavement surface distress by the image information, wavelet neural network(WNN) is put forward to classify sub-image which is made up of 4040 pixels of a pavement surface image. Compare between pattern classifier based on traditional BP artificial neural network and that based on WNN was carried, which proved later one is better than the previous one when other conditions are the same, which provide a new method to exploit more efficient automatic pavement surface distress survey system.

Key words: pavement surface distress survey, image processing; distress recognition; wavelet neural network