Urooj Khan, Samra (2023) Noisy image quality improvement using cobinational filter models and sharpening. Masters thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p SAMRA UROOJ KHAN.pdf Download (596kB) | Preview
	Text (Copyright Declaration) SAMRA UROOJ KHAN COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (558kB) | Request a copy
	Text (Full Text) SAMRA UROOJ KHAN WATERMARK.pdf Restricted to Registered users only Download (7MB) | Request a copy

Abstract

Image is one of the prominent mediums used to illustrate or express a message in daily communication. Image is used in a variety of applications nowadays, such as in security systems, communication systems and medical systems. The feature of an image is its vast data capacity, particularly for high- resolution images. In today's world, noisy image is one worry in developing countries like Pakistan. When working with noisy images, existing methods remove the noise but also the details of the image edges, resulting in a blurred final image. There is no distinction between the image's edges and background. To overcome this problem, combination filter models have been developed so that they can be used to solve various types of noise problems. In this research the noise is introduced to produce a noisy image sample. Then, Mean filters, Median filters, and Wiener filters are used to eliminate noise from the image samples. Next, to detect the edges in the image, the Sobel, Prewitt, Laplacian of Gaussian and Canny edge detection techniques are used. Meanwhile, the Laplacian Operator is applied which, sharpens the blurred edges of an image. All the proposed models were tested using eight sample images. The experimental findings show that Combination Model 1 with Laplacian Operator is effective at removing Salt and Pepper noise with Peak Signal to Noise Ratio value of 40.66. Combinational Model 2 with Laplacian Operator shows good results for Salt and Pepper noise with a 38.89 Peak Signal to Noise Ratio value. While Combinational Model 3 with Laplacian Operator reveals a 36.60 Peak Signal to Noise Ratio value for Poisson noise. Combinational Model 1 and 3 equally give good value for Speckle noise with 34.79 and 34.02 Peak Signal to Noise Ratio. These models give the best value for portrait, landscape and standard-size images. The findings show that the proposed models outperform previously proposed methods in terms of Peak Signal to Noise Ratio quality

Actions (login required)

View Item