基于机器视觉的城市空间尺度要素特征提取系统设计

Design of urban spatial scale elements feature extraction system based on machine vision

针对传统城市空间尺度要素特征提取系统没有分类处理遥感影像多尺度纹理特征,导致提取效果差、精度低等问题,设计一种基于机器视觉的城市空间尺度要素特征提取系统。基于机器视觉技术将数字图像处理与计算机理论相结合,采用灰度共生矩阵法提取遥感影像多尺度纹理特征,并利用决策树分类法对其展开分类后,获取城市遥感影像分类结果。通过直方图阈值分割方法提取水体特征,利用决策树分类法提取建筑物特征,通过纹理参数分割、数字形态学处理、特征矢量化等步骤提取植被特征,运用数学形态学和边缘检测提取道路、阴影以及裸地特征。实验结果表明,该系统空间尺度要素特征提取,准确率高达95.02%,精度高;并且速度快、效率高,说明该系统实用性较好。

In allusion to the problem that traditional urban spatial scale element feature extraction system does not classify and process the multi-scale texture features of remote sensing images,which results in the poor extraction effect and low precision,an urban spatial scale elements feature extraction system based on machine vision is designed.The digital image processing is combined with computer theory based on machine vision technology,the multi-scale texture features of remote sensing images are extracted by means of the gray scale co-existing matrix method,and the remote sensing images are classified by means of the decision tree classification to obtain the classification results of urban remote sensing images.The water features are extracted by means of the histogram threshold segmentation,the building features are extracted by means of the decision tree classification,the vegetation features are extracted by the texture parameter segmentation,digital morphological processing and feature vectorization,and the features of the road,shadow and bare ground are extracted by the mathematical morphology and edge detection.The experimental results show that the extraction accuracy of spatial scale element feature of the system is95.02%,the extraction precision is high,and the extraction efficiency and speed is excellent,which indicates that the system is practical.