Open Access Open Access  Restricted Access Subscription Access

River Basin Extraction from satellite images using Back Proportion Neural Network


Affiliations
1 Department of Computer Science, Balbhim Arts, Science and Commerce College, Beed (M S), India
2 Dr Babasaheb Ambedkar Marathwad University, Aurangabad (M S), India
 

A river network is usually a system wherever in all the tributaries of the rivers, lakes or streams be a part of to create a basin. The basin includes of the laborious and soft rocks that square measure fashioned by the influence of weather conditions, vegetation and transport of sediments and water. The stream network model identity’s the placement of water bodies, determines the causes of floods, deposit, pollution of stream bodies and preventive ways. The pre-processing of the data obtained from the GSI helps to evaluate and investigate the data more accurately and efficiently in predicting the water resources and determining the quality of the water. Some of the problems that are addressed in the proposed research study are caused by the pixel-based indexes leads to an error in the detection of water due to the other occlusions like the cloud shadows and the noise that is incorporated during the image fusion process has to be eliminated for a more enhanced quality of image. The primary aim of the proposed research model are to develop an enhanced multi-temporal pixel level image fusion with advanced image classification technique that detects the changes in the surface of water and demonstrate GIS image segmentation based on convolution wavelet neural network by adding an adaptive filter to further improve the segmentation process. The proposed model will be extended by integrating the other machine learning models to create a hybrid or both can be compared such as SVM (support vector machine), ANN (Artificial neural network) or ML (maximum likelihood) classification. Further, the model is developed by adding filters that completely eliminates the noise and that are more adaptive and robust in nature.

Keywords

Feature Extraction, NDWI, DME, SVM, BPNN.
User
Notifications
Font Size


  • River Basin Extraction from satellite images using Back Proportion Neural Network

Abstract Views: 249  |  PDF Views: 0

Authors

B. Kale Suhas
Department of Computer Science, Balbhim Arts, Science and Commerce College, Beed (M S), India
W. Gawali Bharti
Dr Babasaheb Ambedkar Marathwad University, Aurangabad (M S), India

Abstract


A river network is usually a system wherever in all the tributaries of the rivers, lakes or streams be a part of to create a basin. The basin includes of the laborious and soft rocks that square measure fashioned by the influence of weather conditions, vegetation and transport of sediments and water. The stream network model identity’s the placement of water bodies, determines the causes of floods, deposit, pollution of stream bodies and preventive ways. The pre-processing of the data obtained from the GSI helps to evaluate and investigate the data more accurately and efficiently in predicting the water resources and determining the quality of the water. Some of the problems that are addressed in the proposed research study are caused by the pixel-based indexes leads to an error in the detection of water due to the other occlusions like the cloud shadows and the noise that is incorporated during the image fusion process has to be eliminated for a more enhanced quality of image. The primary aim of the proposed research model are to develop an enhanced multi-temporal pixel level image fusion with advanced image classification technique that detects the changes in the surface of water and demonstrate GIS image segmentation based on convolution wavelet neural network by adding an adaptive filter to further improve the segmentation process. The proposed model will be extended by integrating the other machine learning models to create a hybrid or both can be compared such as SVM (support vector machine), ANN (Artificial neural network) or ML (maximum likelihood) classification. Further, the model is developed by adding filters that completely eliminates the noise and that are more adaptive and robust in nature.

Keywords


Feature Extraction, NDWI, DME, SVM, BPNN.

References