Please use this identifier to cite or link to this item:
Title: Image Restoration Using Adaptive Soft Computing Techniques
Authors: Nadeem, Muhammad
Keywords: Computer Science
Computer & IT
Issue Date: 2020
Publisher: International Islamic University, Islamabad.
Abstract: Generally, real-time images are frequently degraded by various kinds of noise sources. Consequently, succeeding image operations such as image segmentation, object detection and tracking may perform poorly in the presence of noise. So the restoration of noisy images is an active and highly demanded area of research as previous information about noise is almost unknown in many cases making image restoration a more difficult and challenging job. Noise removal is a fundamental step that plays a vital and challenging role in the area of signal and image processing. This research is an attempt to suppress low and high categories of two types of noises i.e., Impulse noise present in almost every image and Speckle noise from ultrasonic data in such a manner to enhance the relevant image content. For this, unsupervised filtering techniques based on statistical and widely used soft computing technique – fuzzy logic, are applied to suppress the impulse and speckle noise hence improving its quality for subsequent image processing operations specifically for diagnostic purposes. In this thesis, the following contributions have been made in the domain of Image Restoration (IR). In the first phase, an efficient image restoration technique, Quadrant based Spatially Adaptive Fuzzy filter based on spatially linked directional adjoining pixels and fuzzy logic for addressing moderate and highly corrupted grayscale images with the challenging type of impulse noise that is Random Valued Impulse Noise (RVIN) is presented. The proposed technique decomposes a larger sized impulsive region of an image into numerous overlapping small patches for the estimation of lower as well as a higher degree of impulse noise, with enhanced image restoration results. In the second phase, an innovative Fuzzy logic based Non-Local Mean filter is introduced in this thesis to model the speckle noise and to restore the degraded image using Fuzzy Uncertainty Modelling (FUM), smoothed by local statistic based information with the capability of retaining the fine details present in the low as well as highly speckled ultrasound images. Objective analysis performed using popular quantitative measures and subjective evaluation of the results show the efficacy of the proposed filters over most of the bench-marked denoising filters in removing impulse as well as speckle noise while retaining the edges and other important details present in the image.
Gov't Doc #: 21951
Appears in Collections:PhD Thesis of All Public / Private Sector Universities / DAIs.

Files in This Item:
File Description SizeFormat 
muhammad nadeem cs 2020 iiui isb.pdfphd.Thesis2.03 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.