الفهرس 
Chapter 1 : IntroductionOnly 14 pages are availabe for public view
 |  | from | 124 |  |  |
| | | from | 124 | | |
Abstract
Although capturing infrared images is simple, perceptual visualization is complex due to environmental factors such as light rain, partially overcast, mostly cloudy skies, haze, inadequate lighting, sensor noise, geographical distance, and item appearances. Image enhancement is a necessary method to improve the human perception and quality of infrared images for further processing such as image analysis. Due to data collection and transmission, infrared images have low contrast and poor image quality. Over enhancing and brightness distortion have been documented while using traditional techniques for infrared image enhancement. To enhancement image quality, this thesis proposed an infrared image enhancement technique based on discrete wavelet transform (DWT) and (DWT) image fusion. The steps of the method are divided into two stages. At the first stage, the infrared image with low contrast is initially presented into the (DWT) algorithm, which divides the image into four matrices. The (DWT) phase is applied again to the (low- low) matrix, which is the matrix that contains most of the details of the image. Thereafter, an update process is performed by Bicubic Interpolation to return those filters to the original image size and then consolidated by (DWT Image Fusion). After these steps, Inverse Discrete Wavelet Transfer (IDWT) is applied. On the other hand, the (low -low) matrix is worked on again by the Laplacian Filter application, resulting in a filtered image. DWT Image Fusion is applied to it with the inversed infrared for an intermediate enhanced image.
In the second phase, the infrared image with low contrast is taken again and image denoising is applied. Then apply the Sobel Operator phase to get an image that is combined with the enhanced intermediate image to get the enhanced (IR) image in its final form using (DWT Image Fusion). A quantitative comparison shows that the proposed technique performs better with an average peak signal-to-noise ratio (PSNR).