Classified satellite images into categories in which they are suitable for- agricultural, commercial, industrial etc. using Image Processing and Machine Learning algorithms.
Geographically, a city is characterized as a patchwork of intensive land-uses. Land-use is the rational and judicious approach of allocating available land resources for different activities (such as settlements, arable fields, pastures, and managed woods) within a city. It is a way of utilizing the land, including the allocation, planning, and management of its resources. The use of a particular patch of land and its physical character are linked. However, research that establishes this link is lacking despite the proliferation of geospatial data. Linking a city’s physical form with its function is the goal of this project.
Land cover comprises of vegetation and resources like grass, asphalt, trees, bare ground, water, etc. Land cover data basically documents how much of a region is covered by forests, wetlands, impervious surfaces, agriculture, and other land and water types (including wetlands or open water). By analyzing satellite and aerial imagery, the land cover can be determined. Identification, delineation and mapping of this land cover establish the baseline from which global monitoring activities like change detection, further studies, resource management, and planning activities can take place. The land cover also provides the ground cover information for baseline thematic maps.
The system consists of five major components namely:
1. Pre-Processed Landsat Images.
2. NDVI Computation.
3. Image Masking.
4. Land-Cover Classification
5. Land-Use Map
Impervious and pervious (vegetation and soil) surfaces can be encoded into numerical categories and classified using machine learning algorithms. In this project, the VIS can be modeled by taking advantage of the linear correlation of impervious and pervious surfaces in very high resolution (0.5mx0.5m pixels) and medium resolution (30mx30m pixels) satellite images. Impervious surfaces can then be further characterized according to their morphology within arbitrarily defined land-use boundaries and classified into land-use categories.
Input is the Land-Cover Map of a city. This image is the training dataset for further classification of different images. This image is input in the ArcGIS software and NDVI computation takes place, followed by image masking and classification to obtain the Land-use map
Output is the Land-Use Map which indicates all the categories.
According to our algorithms, our accuracy is roughly 80%
The goal of this system is to obtain the land-use map which highlights the areas suited to the users’ requirements. This makes construction easier. For example, a user can now use the commercial regions specified in the land-use map for construction of commercial offices.