复杂场景下非正交建筑图像自动识别仿真  

Automatic Recognition Simulation of Non-Orthogonal Building Images in Complex Scenes

作　　者：蔡小青[1] CAI Xiao-qing(City College of Science and Technology,Chongqing University,Chongqing 402167,China)

机构地区：[1]重庆大学城市科技学院

出　　处：《计算机仿真》2019年第10期339-343,共5页Computer Simulation

基　　金：基于全程BIM的跨专业毕业设计教学模式探究与实践(审批单位:重庆市教委)(171030)

摘　　要：针对现有复杂场景下建筑图像自动识别效果差、精度不高的问题,提出基于复杂场景的非正交建筑图像自动识别方法.通过仿射变换的方法对图像进行几何变换,将邻域平均法和中值滤波方法相融合,对经过变换后的建筑图像进行去噪处理,降低图像中存在的噪声.对去噪后的非正交建筑图像进行目标重构,采用逆街距离变换的方法对距离图像轮廓匹配相关数值进行变换,突出目标区域与非目标区域之间的灰度对比度.再对重构图像进行特征提取,结合目标区域残差的形态完成图像目标识别.实验证明,所提方法对复杂场景下非正交建筑图像目标识别的效果较好,提高了图像自动识别的精度.At present,the automatic recognition effect is poor and the recognition precision is low in the complex scene. This paper presented a method to automatically recognize the non-orthogonal architecture image based on complex scenes. Firstly,the image was geometrically transformed by affine transformation method,and then the neighborhood average method and the median filtering method were combined to remove the noise in the transformed building image,so as to reduce the noise existing in the image. Then,the target reconstruction was performed on the non - orthogonal architecture image after the noise reduction. Moreover,the inverse city block distance transform method was used to change the correlation value of distance image contour matching,and the gray contrast between the target area and the non - target area was highlighted. The features of reconstructed image were extracted again. Combined with the shape of target region residual,the image target recognition was completed. Simulation results show that the proposed method has better effect on target recognition of non - orthogonal building image in complex scenes. Meanwhile,this method improves the accuracy of automatic recognition.

关 键 词：自动识别 非正交建筑图像 复杂场景 中值滤波 邻域平均法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]