Current Science

Open Access Open Access  Restricted Access Subscription Access

Perspective Forward Enhancement in Boundaries of Satellite Image via BO-TDyWT for Precise Delineation and Accurate Measurement


Affiliations
1 Department of Artificial Intelligence and Data Science, Aalim Muhammed Salegh College of Engineering, Chennai 600 054, India
2 Department of Computer Science Engineering, Aalim Muhammed Salegh College of Engineering, Chennai 600 054, India
3 Institute of Remote Sensing, Anna University, Chennai 600 025, India

Delineation of vegetation and water body regions on the earth’s surface plays a vital role in the development and planning of an area. For delineation, traditional algorithms and classifiers require more training datasets and interpretation time. For accurate delineation, geometric distortion and mixed pixels in satellite images need to be removed. Geometric distortion is due to various factors such as relief displacement, variations in the satellite altitude and attitude, and curvature of the earth’s surface. Mixed pixels arise due to different types of land cover in an area. In this study, the Bayesian optimized transverse dyadic wavelet transform (BO-TDyWT) algorithm enhances the edges and curvatures of a region in an image. BO-TDyWT classifies the vegetation and water bodies in LANDSAT image, which consists of different terrains such as hilly, land and coastal regions. BO-TDyWT removes the geometric distortion and mixed pixels in the hill area water body. Performance of proposed BO-TDyWT algorithm is compared with dyadic wavelet transform and TDyWT. From the results, BO-TDyWT accurately delineates hill areas, vegetation and water body areas than dyadic wavelet transforms and TDyWT. BO-TDyWT results are ground truth verified. The BO-TDyWT algorithm accurately delineates vegetation and water bodies for precise measurement with an accuracy of 96%, which is higher than TDyWT.

Keywords

Dyadic wavelet transform, geometric distortion, mixed pixels, satellite images.
User
Notifications
Font Size

Abstract Views: 63




  • Perspective Forward Enhancement in Boundaries of Satellite Image via BO-TDyWT for Precise Delineation and Accurate Measurement

Abstract Views: 63  | 

Authors

M. Prabu
Department of Artificial Intelligence and Data Science, Aalim Muhammed Salegh College of Engineering, Chennai 600 054, India
N. R. Shanker
Department of Computer Science Engineering, Aalim Muhammed Salegh College of Engineering, Chennai 600 054, India
K. Srinivasan
Institute of Remote Sensing, Anna University, Chennai 600 025, India

Abstract


Delineation of vegetation and water body regions on the earth’s surface plays a vital role in the development and planning of an area. For delineation, traditional algorithms and classifiers require more training datasets and interpretation time. For accurate delineation, geometric distortion and mixed pixels in satellite images need to be removed. Geometric distortion is due to various factors such as relief displacement, variations in the satellite altitude and attitude, and curvature of the earth’s surface. Mixed pixels arise due to different types of land cover in an area. In this study, the Bayesian optimized transverse dyadic wavelet transform (BO-TDyWT) algorithm enhances the edges and curvatures of a region in an image. BO-TDyWT classifies the vegetation and water bodies in LANDSAT image, which consists of different terrains such as hilly, land and coastal regions. BO-TDyWT removes the geometric distortion and mixed pixels in the hill area water body. Performance of proposed BO-TDyWT algorithm is compared with dyadic wavelet transform and TDyWT. From the results, BO-TDyWT accurately delineates hill areas, vegetation and water body areas than dyadic wavelet transforms and TDyWT. BO-TDyWT results are ground truth verified. The BO-TDyWT algorithm accurately delineates vegetation and water bodies for precise measurement with an accuracy of 96%, which is higher than TDyWT.

Keywords


Dyadic wavelet transform, geometric distortion, mixed pixels, satellite images.



DOI: https://doi.org/10.18520/cs%2Fv127%2Fi5%2F560-571