Date of Submission


Date of Award


Institute Name (Publisher)

Indian Statistical Institute

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Subject Name

Computer Science


Electronics and Communication Sciences Unit (ECSU-Kolkata)


Chanda, Bhabatosh (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

A scene to be photographed, usually includes objects at varying distances from the camera. Depth-of-field of a digital camera is the range of distance, all objects within which appear to be sharp in the image. Due to the low depth-of-field of the camera, images acquired by them often suffer from degradation called out-of-foc us blur. One way to enhance the effective depth-of-field is to acquire se veral images of a scene with focus on different parts of it and then combine these images into a single image in such a way that all regions of the scene are in focus. Acquir ed images are called multi- focus ima ges and the process of combination is known as multi- focus image fusion. The techniques for multi-foc us image fusion belong to the broad categories, pi xel-based, block-based and region-based. They concentrate respectively on single pixels, small blocks of size m xn and arbitrarily shaped regions. Image registration is a necessary pre-requisite for multi-focus image fusion. The thesis presents a new technique for multi-focus image registration and three new techniques for multi-focus image fusion. Among these techniques, the first one is pixel-based, the second one is block-based and the third one is region-based. All of them use mathematical mo phological tools. The pixel-based method is a multi-resolu tion technique that employs mor phdogical wavelet as a tool for signal decomposition and reconstruction. The block-based method uses energy of morphological gradients as a tool for focus measure. Finally the region-based method uses multi-scale morphological tools for obtaining the focu sed regions from the input images. In this context, existing fusion tech niques are studied and categorized. The thesis includes experimen tal results obtained by applying the proposed methods and ot her well-known methods on a variety of input data-set. It also includes performance analysis of various methods using standard quan ti tative e valuation techniques. At the end it presents concluding remarks and a discussion on related future work.


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


Included in

Mathematics Commons