无人机降落地点的智能障碍图像识别方法仿真  

Unmanned Aerial Vehicle(UAV) Intelligent Obstacle Image Recognition Method for the Simulation of Landing Site

作　　者：谢旻旻[1] 

机构地区：[1]青海民族大学计算机学院,青海西宁810007

出　　处：《计算机仿真》2015年第7期84-87,共4页Computer Simulation

基　　金：面向青海土族民俗数字化的互动展示系统(2014-GX-C09)

摘　　要：无人机在降落过程中,对降落地点中的障碍图像进行准确识别能够有效地保证无人机的安全。利用传统算法进行障碍图像识别的过程中,一旦障碍物与周围环境的特征差异较小,将会造成识别准确率降低的问题。为解决上述问题,提出一种采用支持向量机的无人机降落地点的智能障碍图像识别方法。首先利用降落地点图像中的R、G、B分量的灰度值的前三阶矩作为障碍图像颜色的特征并进行提取,然后利用对比度、相关性、能量、同质性和维度这五个参数对障碍图像的纹理特征进行提取,最后使用支持向量机来最终完成对障碍图像的准确识别。仿真结果表明,使用改进算法进行对无人机降落地点的障碍图像进行仿真识别,能够提高识别的准确性,同时也提高了识别的效率。Unmanned aerial vehicle（ uav） in the process of landing,the landing site of the obstacle image accurately recognition can effectively guarantee the safety of the unmanned aerial vehicle（ uav）. The use of traditional algorithm for obstacles in the process of image recognition,once the obstacles and the characteristics of the surrounding environment difference is small,will cause problem identification accuracy. For this,put forward a kind of unmanned aerial vehicle（ uav） landing site based on support vector machine（ SVM） method of intelligent obstacle image recognition. First using the landing point in the image grey value of R,G,B component as a barrier in the first three moments of the image color feature extraction,and then use contrast,correlation,energy,homogeneity and dimensions of the five parameters on the obstacle image texture feature is extracted,finally using support vector machine（ SVM）to the final obstacle to the accurate recognition of the image. Simulation experiment results show that using the improved algorithm for uav landing site obstacle image recognition,can improve the accuracy of recognition,but also improve the efficiency of recognition.

关 键 词：无人机 降落地点 图像识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]