Research Description

Underwater navigation and intelligent object recognition require robust machine learning algorithms to operate in turbid water. Modern life created the man-made pollution in oceans, rivers, and lakes, which contaminate our water resources. The underwater vehicle can be used for survey purposes. Submersible captured images are in degradation quality due to the underwater environment. To overcome this, Enhancement experimental analyses are carried out for both Image Formation Model (IFM) - free and IFM-based techniques. It is observed that proposed visibility improvement techniques have shown overwhelming qualitative observation. For the classification of submerged objects in the captured imagery, the Convolutional neural Network (ConvNet) is trained. Finally, the feature map size is condensed and complex features are extracted and passed into a neural network. The output of the network results in a probability distribution over L classes. Stochastic gradient descent with ADAM optimizer is used to square the gradients in order to scale the learning rate and reduces the difference between the actual and predicted classes. It is found that the proposed ConvNet underwater image recognition detector, achieves good accuracy than the existing multiclass detector.

Publications:

Journals:

1. Dhana Lakshmi Manikandan, Sakthivel Murugan Santhanam, "Parallel desires: unifying local and semantic feature representations in marine species images for classification", Marine Geophysical Research, ISSN 1573-0581, vol. 45:16, pp. 1-30, July 2024. 

2. Dhana Lakshmi MANIKANDAN, Sakthivel Murugan SANTHANAM, "Underwater species classification using deep learning technique", Romanian Journal of Information Technology and Automatic Control, ISSN 1220-1758, vol. 34(2), pp. 7-20, June 2024.

3. Dhana Lakshmi M, Sakthivel Murugan S, “Contrast Improvement on Side Scan Sonar images using Retinex based edge preserved technique", Marine Geophysical Research, Springer, May 2022. 

4. Dhana Lakshmi M, Sakthivel Murugan S, “Visibility Improvement of Underwater Turbid Image using Hybrid restoration network with Weighted filter”, Multidimensional System and Signal Processing", Springer, Nov., 2021. 

5. Dhana Lakshmi M, Sakthivel Murugan S, ‘Modified Restoration technique for improved visual perception of shallow underwater imagery’, Current Science, Vol. 121, Issue 1, July 2021, pp. 103-108. 

6. Dhana Lakshmi M, Sakthivel Murugan S, “Keypoint-Based Mapping Analysis on Transformed Side Scan Sonar Images”, Multimedia Tools and Applications 79, 26703–26733, Springer, 2020.

Conferences:

1. Dhana Lakshmi M, Vimal Raj M, Sakthivel Murugan S, “Intensity Based Shadow Classification- SSS Images”, National Conference on Challenges in Earth System Sciences for Global Sustainability (CESS-GS), held at IIT Kharagpur between Jan 15-17, 2020, India 

2. Dhana Lakshmi M, Sakthivel Murugan S, “Underwater Image Recognition Detector using Deep ConvNet”, Twenty Sixth National Conference on Communications (NCC 2020), held at IIT Kharagpur, India, between Feb 21-23, 2020. 

3. Dhana Lakshmi M, Sakthivel Murugan S, Padmapriya N, Somasekar M, “ Texture Analysis on Side Scan Sonar images using EMD, XCS-LBP and Statistical Co-Occurence” in the proceedings of 2019 International Symposium on Ocean Technology (SYMPOL 2019), held at Department of Electronics, Cochin University of Science and Technology (CUSAT), Cochin, India, between 11-13, Dec 2019. 

4. Dhana Lakshmi M, Vimal raj M, Sakthivel Murugan S, “Feature Extraction and Matching on geometrically distorted Side Scan Sonar images”, International Conference on Microwave Integrated Circuits, Photonics and Wireless Networks (IMICPW-2019), May 22nd – 24th, 2019, NIT Trichy, Trichy.