Open Access Open Access  Restricted Access Subscription Access

Advancing Gis Map Maintenance: Change Detection And Update Using ResU-Net: A Case Study on Chandigarh and Hyderabad Cities, India


Affiliations
1 Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada 520 007, India
2 National Remote Sensing Centre, ISRO, Hyderabad 500 037, India

India’s metropolitan cities have been growing rapidly for many years. To keep geographical information accurate and current, it is essential to update GIS maps. Traditionally, experts have analysed new data sources and made necessary adjustments to the maps manually. Such manual monitoring is a laborious test both economically and in terms of workforce. Geographical data are transformed into digital maps by GIS mapping, making it simple to spot patterns, trends and linkages. Extraction of humanmade objects, such as roads, water bodies and buildings, from remotely sensed imageries holds significance in various urban applications, including urban land-use and land-cover assessment, geographical database updates and change detection. Cartosat-3 data can provide detailed information about buildings and their changes over time. Additionally, GIS maps are manually updated by rasterizing vector data. The suggested system consists of ResNet and U-Net architecture as its core. The bi-temporal images are initially co-registered to completely align 2020 and 2022 satellite images with respect to the coordinates. Buildings are then segmented using U-Net with ResNet as the backbone, and the resultant segments are converted from raster to vector format. The suggested model has been tested and trained using the Chandigarh dataset, which resulted in an accuracy of 95%.

Keywords

Change detection, digital maps, geographical data, remote sensing, urban planning
User
Notifications
Font Size

Abstract Views: 139




  • Advancing Gis Map Maintenance: Change Detection And Update Using ResU-Net: A Case Study on Chandigarh and Hyderabad Cities, India

Abstract Views: 139  | 

Authors

S. Vasavi
Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada 520 007, India
Dibyaranjan Parida
Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada 520 007, India
N. Malathi
Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada 520 007, India
M. Sobhana
Velagapudi Ramakrishna Siddhartha Engineering College, Vijayawada 520 007, India
Reedhi Shukla
National Remote Sensing Centre, ISRO, Hyderabad 500 037, India

Abstract


India’s metropolitan cities have been growing rapidly for many years. To keep geographical information accurate and current, it is essential to update GIS maps. Traditionally, experts have analysed new data sources and made necessary adjustments to the maps manually. Such manual monitoring is a laborious test both economically and in terms of workforce. Geographical data are transformed into digital maps by GIS mapping, making it simple to spot patterns, trends and linkages. Extraction of humanmade objects, such as roads, water bodies and buildings, from remotely sensed imageries holds significance in various urban applications, including urban land-use and land-cover assessment, geographical database updates and change detection. Cartosat-3 data can provide detailed information about buildings and their changes over time. Additionally, GIS maps are manually updated by rasterizing vector data. The suggested system consists of ResNet and U-Net architecture as its core. The bi-temporal images are initially co-registered to completely align 2020 and 2022 satellite images with respect to the coordinates. Buildings are then segmented using U-Net with ResNet as the backbone, and the resultant segments are converted from raster to vector format. The suggested model has been tested and trained using the Chandigarh dataset, which resulted in an accuracy of 95%.

Keywords


Change detection, digital maps, geographical data, remote sensing, urban planning



DOI: https://doi.org/10.18520/cs%2Fv126%2Fi3%2F326-335