Enhancement and Restoration of Microscopic Images Corrupted with Poisson's Noise Using a Nonlinear Partial Differential Equation-based Filter

Rajeev Srivastava; JRP Gupta; Harish Parthasarathy

doi:10.14429/dsj.61.1181

Rajeev Srivastava IT BHU
JRP Gupta Netaji Subhas Institute of Technology, New Delhi
Harish Parthasarathy Netaji Subhas Institute of Technology, New Delhi

DOI: https://doi.org/10.14429/dsj.61.1181

Keywords: Image restoration, Poisson noise, fluorescence microscopic images, PDE based nonlinear filter, MAP, likelihood, regularisation

Abstract

An inherent characteristic of the many imaging modalities such as fluorescence microscopy and other microscopic modalities is the presence of intrinsic Poisson noise that may lead to degradation of the captured image during its formation. A nonlinear complex diffusion-based filter adapted to Poisson noise is proposed in this paper to restore and enhance the degraded microscopic images captured by imaging devices having photon limited light detectors. The proposed filter is based on a maximum a posterior approach to the image reconstruction problem. The formulation of the filtering problem as maximisation of a posterior is useful because it allows one to incorporate the Poisson likelihood term as a data attachment which can be added to an image prior model. Here, the Gibb's image prior model-based on energy functional defined in terms of gradient norm of the image is used. The performance of the proposed scheme has been compared with other standard techniques available in literature such as Wiener filter, regularised filter, Lucy-Richardson filter and another proposed nonlinear anisotropic diffusion-based filter in terms of mean square error, peak signal-to-noise ratio, correlation parameter and mean structure similarity index map.The results shows that the proposed complex diffusion-based filter adapted to Poisson noise performs better in comparison to other filters and is better choice for reduction of intrinsic Poisson noise from the digital microscopic images and it is also well capable of preserving edges and radiometric information such as luminance and contrast of the restored image.

Defence Science Journal, 2011, 61(5), pp.452-461, DOI:http://dx.doi.org/10.14429/dsj.61.1181

Author Biographies

Rajeev Srivastava, IT BHU

Dr Rajeev Srivastava received his ME (Computer Technology and Applications) and PhD (Computer Engg) both from University of Delhi, India, in 2005 and 2011 respectively. Currently working as an Associate Professor in the Department of Computer Engineering, Institute of Technology, Banaras Hindu University (IT-BHU), Varanasi, India. He has 26 research publications in national/ international conferences and journals and 04 book chapters to his credit. His research interests include image processing and vision, and algorithms.

JRP Gupta, Netaji Subhas Institute of Technology, New Delhi

Dr J.R.P. Gupta received his BSc(Electrical Engineering) and PhD from the University of Bihar, India, in 1972 and 1983 respectively. Curently working as a Head, Department of Instrumentation and Control Engineering, University of Delhi. He is a senior member of IEEE and recipient of IETE K.S. Krishnan memorial award for the best system oriented paper. His research interest includes signal processing, power electronics and control systems.

Harish Parthasarathy, Netaji Subhas Institute of Technology, New Delhi

Dr Harish Parthasarthy received his BTech (Electrical Engineering) from Indian Institute of Technology (IIT), Kanpur, India, in 1990 and PhD from (IIT), Delhi, India, in 1994. He has also pursued Post-doctorate for a period of one year at Indian Institute of Astrophysics (IIAP), Bengaluru, India. Currently he is a Professor in the Department of Electronics and Communication Engineering, Netaji Subhas Institute of Technology (NSIT), New Delhi, India. His research interest include signal processing.