Research And Development Of Urban New Construction Land Change Detection Based On High-resolution Image

The development of urbanization in my country is changing with each passing day.Real-time monitoring of urban construction land changes is particularly important for the country's macro-control and scientific governance.The detection of urban newly-added construction land changes helps city managers to effectively grasp the status of urban land changes,rationally plan urban development,and reduce the occupation of agricultural land and illegal construction.With the continuous development of high spatial resolution and high temporal resolution remote sensing technology,the use of remote sensing images to develop object-oriented methods for detecting changes in land surface coverage is becoming more and more popular.To this end,the article uses the high-resolution remote sensing images of the previous two phases to study the change detection of urban new construction land.The data used in this article are remote sensing images of GF-1 and GF-2 and the results of the previous geography and national conditions monitoring vector.After the remote sensing image data is correspondingly preprocessed,the ground feature image pattern is obtained by superimposing the results of the previous geographic national conditions monitoring vector and performing vector constraint segmentation;on this basis,the spectral and texture features of the ground feature are extracted to construct a feature vector,and the vector is similar To determine whether the corresponding area changes in the early and late stages;finally,the false changes are further removed from the detection results to obtain the final change detection results.The main research contents of this paper are as follows:(1)Remote sensing image segmentation with vector constraints.Remote sensing image segmentation is the first step of the object-oriented change detection method.The result of segmentation will affect the accuracy of change detection to a certain extent.In this paper,we use the boundary of the vector results of the previous geography and national conditions to perform vector constrained segmentation of the later remote sensing image,which limits the occurrence of over-segmentation and ensures the accuracy of image segmentation.Then,the segmentation result is mapped to the previous image to obtain the segmented image spots that completely correspond to the previous and late images.(2)Change detection technology based on vector similarity.After obtaining the same segmentation patterns in the early and late stages,extract the spectral and texture features of the segmented patterns in the early and late stages,analyze the feature changes of the changed patterns,and select several groups of features with large changes to construct feature vectors.Then calculate the vector similarity between the feature vectors in the early and late stages,and judge the changes according to the given threshold.At the same time,in order to ensure that the detected results are all types of construction land,the article constructs the Difference Construction Land Index(DCLI)to remove pseudo-change patterns and improve the accuracy of change detection.(3)Research and development of software for detecting changes in urban newly-added construction land.The traditional change detection method mainly relies on manual visual interpretation,which has a large workload and low efficiency.Therefore,on the basis of(1)and(2),based on the domestic secondary development component set PIE-SDK,combined with the Visual Studio 2017 development platform,the development of urban new construction land change detection tool software is carried out.It realizes functions such as loading of raster and vector data,vector constraint segmentation,boundary mapping,feature extraction and vector similarity calculation.(4)Tool software experiment and analysis.Taking some areas of Chongqing as an example,this paper uses the newly developed construction land change detection tool software to carry out change detection experiments.Through several experiments,appropriate image segmentation parameters and change threshold range are selected,and the experimental results can reach the rate of missed detection within 10% and the accuracy rate is as high as 90%.The experimental results show that the method presented in this paper is effective and the tool software has a certain practical value.In summary,on the basis of obtaining the same segmentation spots before and after the period,the paper uses the vector similarity method to detect the change of urban new construction land,and removes the false change of the change detection results,which can effectively reduce the false detection rate.The development of new construction land change detection tool software can reduce the workload of manual extraction of change areas and improve the efficiency and automation of change detection.