Extracting Urban Building Information Based On ADS100 Images

According to the platform of the sensor,the high-resolution remote sensing images mainly include high-resolution satellite remote sensing images,high-resolution digital aerial photography remote sensing images,and high-resolution land-based remote sensing images.Information about buildings,shadows,vegetation,roads,and water features is common in the above images.The types and divisions of urban buildings reflect local political,economic,and demographic information,as well as national characteristics and local customs.Their related research results are important in national defense,disaster prevention,urban planning and development,and economic construction.effect.However,the extraction of urban building information is susceptible to interference from roads,bare soil,concrete floors,etc.Therefore,how to efficiently and accurately extract buildings has become the focus and difficulty of industry research.ADS-100 digital aerial camera is equipped with SH120 type lens,CCD pixel can reach 20,000 pixel value,the image includes red light band,blue light band,green light band and near infrared light band.After consulting related research literatures,it was found that there are few related researches based on ADS-100 image information extraction,and no urban building extraction research using this image has been found.Therefore,on the premise of combining ADS100 image characteristics,building characteristics,and feature pop relations,this paper proposes three methods for extracting urban buildings from ADS-100 images,so as to complete accurate and stable digital aerial photography.,Convenient,and universal urban building extraction technology and model research to promote the more widespread use of digital aerial photography in the field of ground information interpretation.The main research results are as follows:(1)A method for extracting urban buildings based on principal component transformation and multi-band combination for ADS-100 is proposedFirst,the principal component transformation and information enhancement are performed on the multispectral data of the ADS-100 image,and the first component ratio calculation(PC1 / B)and difference calculation(NIR-B)of the principal component are constructed.Second,the multimodal histogram threshold method is used for segmentation.To obtain two binary images of the mixed buildings;finally,perform a logical AND operation on the two mixed images to achieve the purpose of removing disturbing features and obtain the urban building information.(2)A method for extracting urban buildings based on HSV color space transformation for ADS-100First,the ADS-100 image was converted from RGB color space to HSV color space,and the luminance component V was thresholded to obtain the mixed building area S1.Second,the ADS-100 image was used to establish a ratio difference model using the near infrared and blue light bands.The peak histogram threshold segmentation algorithm performs threshold segmentation on it to obtain the mixed area S2;finally,it performs logical AND operation on S1 and S2 to obtain the complete building area.(3)A multi-scale segmentation method for integrated feature components of ADS-100 is proposedFirst,based on a full understanding of the ADS-100 image and the typical feature spectral information in the image,the image enhancement processing is used to construct the luminance feature component,(NIR-B)/(NIR + B)feature component,and the normalized vegetation index NDVI feature.Components to enhance building information;secondly,reconstruct the constructed feature components to obtain a threeband new image;then use ENVI 5.5 software and multi-scale segmentation objectoriented information extraction technology to select the optimal segmentation scale and establish segmentation objects The rule set performs information extraction;finally,the accuracy of the extraction results is checked and verified.A large number of experiments have shown that the three methods proposed in the paper can not only effectively remove the influence of bare soil,hardened roads,and concrete floors,but also detect building information in ADS100 images more accurately.Verified by comparison.It is found that the ADS-100 city building information extraction method based on feature components is the best.