基于BP神经网络的探地雷达图像识别技术

探地雷达作为一种常用的物探方法,在工程和科研方面都具有广泛的应用,在应用中由于地下介质的复杂性,常使得图像的反射原因较难判断,对于地下介质的典型特征也较难统计.为了更好的解译以及快速准确的分析统计出雷达反射图像中的各类特征,本文利用神经网络的高度模糊识别能力,将标准雷达图像输入到神经网络模型中,并通过一系列的改进,提高了网络性能.通过仿真试验表明,改进后的神经网络可以较为高效的对雷达图像中的典型反射特征进行识别与统计,具有很好的应用前景.

Ground Penetrating Radar Image Recognition Algorithm Based on BP Neural Network Algorithm

Ground-penetrating radar is widely applied to study commonly geophysical method.Due to complexity of subsurface features,the potentiality of the technique was quickly recognized by engineering and scientific researchers,which often makes the image more complex and diverse.In order to interpret and analyze the various types of radar reflectivity characteristics image quickly and accurately.In this study,we use fuzzy neural network to identify the standard radar image.Through a series of improvements and adjust parameters,the parameters of the network-based data selection problem was discussed,learning speed was significantly improved.The experimental results show that the recognition rate has improved significantly;convergence rate of BP network was improved and higher recognition rate in effect was got.