An optimized deep network based soil texture prediction and vegetation analysis using satellite image database. (2nd January 2023)

• Record Type: Journal Article
• Authors: Anand, Mallekedi; Jain, Anuj; Kumar Shukla, Manoj
• Is Part Of: Concurrency and computation. Volume 35:Number 7(2023)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2023-01-02
• Subjects: biological standards -- satellite image -- soil textures -- suitable crops -- Tarakeswar region; Parallel processing (Electronic computers) -- Periodicals; Parallel computers -- Periodicals; 004.35
• DOI: 10.1002/cpe.7611
• Languages: English
• ISSNs: 1532-0626

Abstract

Summary: The usage of satellite images has grown rapidly in digital applications. Predicting features in the satellite images using conventional techniques is not an easy task as the images captured from satellites are more complex with highly noisy features. Hence, traditional imaging algorithms cannot analyze the image features and specify the soil textures. In this work, the Tarakeswar satellite images were taken to estimate the soil textures and crops yielding. Hence, to find the soil's chemical properties and suitable crops, the novel bat‐based U‐Net feature prediction system (BUFPS) was developed with the required features. The imported satellite images were initially filtered and entered into the classification phase to forecast the present chemical features and suitable crops in specific soils. After detecting the chemical features, soil textures like sandy, silt, and clay were categorized; the crops like jute, rice, lentil, and potato were considered. Finally, the planned model is executed in the MATLAB environment and has gained outstanding results by achieving the lowest error rate of 8% and a high prediction score of 92%. The proposed model has achieved exact soil texture analysis for complex satellite images. The gained texture analysis outcome is quite better than other existing models.