Restoration scheme for deblurring of images in the presence of impulsive noise

Abstract

Restoration of the image is a significant concern in advanced image processing technology. Images are frequently degraded at some stage of the data acquisition process. The degradation may possibly involve newlineblurring, detail loss owing to sampling, quantization property and a variety of newlinesources of noise. Images and videos have become an integral part of common newlineman as most of the documents dealt in real time are through text, audio, newlineimage or video. The human visual perception is trained by nature to visualize newlinea good quality image or video. Hence image restoration plays an important newlinerole in several image processing applications. The aspiration of image newlinedeblurring is, to get a better and sharp image from a blurry surveillance. Blur newlineis an outward appearance of a reduction in bandwidth of the image due to a newlinedefective image development procedure. It may be because of the relative newlinemotion between the capturing device and the original image. In general, an newlineimage may be corrupted by means of low-pass filters, as they are accustomed newlineto blur or smooth the image. In real time applications, the images acquired newlinefrom acquisition devices tend to be blurred with noise corruption. newlineThis dissertation work focuses on the development of algorithms newlinefor the removal of blur in the presence of impulse noise from digital images newlinein the frequency domain. Various issues related to this problem are studied newlinefor monochrome as well as color images and suitable filtering and hybrid newlinemulti-resolution methods have been suggested and experimented. Fixed valued and Random-valued impulse noises have been dealt with in this thesis with a multi-level median filter used for fixed impulse noise while a novel global threshold vector outlyingness ratio filter has been implemented for newlineremoval of random impulse noise. newline newline