Optimization-Enhanced Microscopic Image Analysis for Accurate Monitoring of Pond Ecosystem Parameters

Abstract

Introduction: Accurately monitoring pH, dissolved oxygen (DO), Alkalinity, Ammonia and Nitrite levels in aquatic environments is vital for maintaining water quality and ensuring ecosystem balance.

Problem: Traditional methods of measuring pH, DO, Alkalinity, Ammonia and Nitrite often involve invasive and time-consuming direct chemical sampling, limiting their effectiveness in monitoring dynamic aquatic environments.

Method: This study introduces an innovative technique for detecting parameters in pond water by analyzing images taken from three distinct water layers: Epilimnion, Metalimnion (Thermocline), and Hypolimnion. These images were captured using an underwater microscopic camera and optimized using three different optimization algorithms such as Battle Royale Optimization (BRO), Gazelle Optimization (GOA), and Brown-bear Optimization (BOA). The optimized images were processed using the Denoising Convolutional Neural Network (DnCNN), integrated with each optimization algorithm(BRO-DnCNN, GOA-DnCNN, and BOA-DnCNN) to effectively denoise and improve the clarity of the images.

Result: This process yielded statistical values such as mean, standard deviation, and other metrics, which were then used to estimate parameters levels of the pond water. Among the three algorithmic combinations, the BRO-DnCNN algorithm outperformed the others, achieving an accuracy rate of approximately 99%making it the most reliable approach for detecting pH,DO,Alkalinity,Ammonia and Nitrite levels.

Significance: The integration of optimization algorithms with DnCNN for image processing proved to be a highly effective method for real-time monitoring of critical water quality parameters.

Conclusion: This technique not only enhances accuracy but also offers a robust framework for environmental monitoring, making it a valuable tool for aquatic ecosystem management.