Research On Building Detection In Remote Sensing Image

Performing building detection and identification by remote sensing image is a significant study topic at present,which has been widely applied in military and commercial field.The ratio of building has reached over 80% in urban areas.Meanwhile,building detection can also offer help to other object's detection,and it's instructive.Therefore,study of remote sensing image in building detection is particularly meaningful.As one of the most common deep learning approach,Convolutional Neural Network(CNN)has been applied into various computer vision and pattern recognition task,verified to boast a good performance.This paper introduces study reality of remote sensing image in building detection and related study method.Take advantage of Harris point to collect local characteristics,and utilize Convolutional Neural Network(CNN)to train and recognize building local characteristics in conrner neighbor.It mainly introduces the improvement for the extract of Harris point and the application on Convolutional Neural Network(CNN).Specific study includes two respects as follows in this paper:First,pretreat remote sensing image,and propose a optimized Harris detection algorithm.This algorithm extracts precise building point information by modified response function and threshold iteration,and selects the set of points by certain ways,and utilize points information to obtain local characteristics by neighbor collection.Then,detect the building by characteristics of local areas as per the formula in the paper,and use Convolutional Neural Network(CNN)to extract characteristics of local areas by learning training of local characteristics.Thus,achieve a good detection ability and also it's very effective.The analytical experimental results turn out that the algorithm can achieve building detection by remote sensing image.To prove the efficiency of the detecting algorithm in the paper,perform a contrast experiment by BP neural network and RCNN.The experimental results shows that the building detecting algorithm by combining Harris point for local characteristics is particularly precise,and it has a certain value for study of remote sensing image in building detection.