Extracting Urban Green Land From High Resolution Remotely Sensed Image In Typical Area Of Nanjing

This study is an application of high-resolution remotely sensored data in extracting urban green landscape. In particular, the capabilities of the data are assessed. The study area was in Nanjing, Jiangsu Province, China. The type and area of urban green land was extracted finely for the high resolution of QuickBird image. An improved traditional classification method was developed to extract the urban vegetation more accurately.●QuickBird image and unsurprised classification methods were used to detect the shadow of remotely sensed imagery and enhanced infrared value helped to extract the vegetation pixel under the shadow.●NDVI and spectrums value were combined to extract urban vegetation, including the vegetation type and area.●GEMI was developed to improve the accuracy of extracting the urban vegetation, and the new equation called Modified Global Environment Monitoring Index.Remotely sensed data have been used for urban land cover mapping for decades. The recent availability of high spatial resolution imagery from satellite sensors such as QuickBird provides new opportunities for detailed urban land cover mapping at very fine scales. The low and decreasing cost of High Resolution (HR) Remotely Sensed satellite imagery such as QuickBird may motivate planning departments, saving considerable time and money.Accurate and detailed land cover information of urban areas is essential for many purposes such as urban land management, urban planning, and urban landscape pattern analysis, environmental study, and others. Thus, the accurate and detailed extraction of urban green land is of great significance. The research content and results were as follows(1) Detecting the shadows and recovering the vegetation under shadow.The problem of shadowing is a challenge raised in digital image processing. Shadows in a remotely sensed imagery occur when objects totally or partially occlude the direct light from a source of illumination, which include shadows cast on the ground feature by high-rise objects, and self shadows. The effect of shadows in remote sensoring has been an important issue. Great difficulty arises in classification and interpretation of shaded objects in an image because of the reduction or total loss of spectral information of those shaded objects.In the monitoring of urban vegetation, the problem of shadowing is particularly significant in high spatial resolution imagery. With the dominance of elevated objects such as buildings, bridges, towers and trees in the landscape, the proportion of the imagery that is affected by shadowing could be relatively large.Shadow detection and removal were investigated in remotely sensed imagery. Shadow detection is the process of identifying the shaded pixels in remotely sensed imagery, in the study unsupervised classification used to detect the shadowing area, whereas a new algorithm applied for restoration shadow and recover the vegetation pixels, the radiometric enhancement methods increased the value of infrared spectrum to be and add to the unsupervised classification image to detected the missing vegetation under the high building shadow.The new application algorithm is the value of Unsupervised classification+(08*Infrared).With raster attribute editor shows the vegetation pixels as trees under the shadow. Shadow removal is used to restore the spectral information of the shaded areas to obtain a shadow-free image.(2) Extract urban vegetation used NDVI and spectrums value.When we derived NDVI and got a NDVI value greater than0.036, a huge number of the pixels marking the vegetation is seen, and it contained the majority of vegetation. After collecting the fraction of the remaining vegetation we get a NDVI value greater than-0.2, but in this case the extracted vegetation cover also included some blue and white metal roofs, which have similar spectral characteristics to vegetation in the red and near-infrared band.To remove this influence of the roofs, the investigation focused on the value of the three features, blue roofs, white metal roofs, and vegetations in the four bands of QuickBird image. The vegetation value in the blue band shows it should be less than380.There was still a problem to extract the pine fir trees which have the similar spectral characteristics to water and artificial playgrounds. There was still a problem to extract the pine fir trees after the application of NDVI>-0.2and Blue band<380. But when we have the value for these three features, we can recognize the pine fir trees in the red band with the value between200and300.(3) Developing Global Environmental Monitoring Index (GEMI) toimprove the accuracy of extraction of urban vegetationGlobal Environmental Monitoring Index (GEMI) has been applied in the study area. It gives good results, but also creates some confusion because of the similar spectral characteristics of blue roofs and vegetation.Global Environmental Monitoring Index (GEMI) has been modified to eliminate and remove the blue roofs influence, which have similar spectral characteristics to vegetation, the factor of blue band as Blue band<380added to (GEMI) equation, then the new equation called Modified Global Environmental Monitoring Index (MGEMI). After applying the MGEMI and obtaining the preliminary green space extraction, it seems that most of the effects of the blue roofs have been eliminated. The use of Modified Global Environment Monitoring Index (MGEMI) was compared to the Global Environment Monitoring Index (GEMI) and NDVI in this research for extracting urban green land and was found to be more accurate.