Application Of Second-Order Histograms In Noise Reduction And Contrast Measurement Of Intensity Images
This research focuses on a type of information provided by the pixels in intensity images. The types of images that are the focus of this research are scanning electron microscopy (SEM) and magnetic resonance imaging (MRI) images. These two types of images were chosen for the amount of details that...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2019
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This research focuses on a type of information provided by the pixels in intensity images. The types of images that are the focus of this research are scanning electron microscopy (SEM) and magnetic resonance imaging (MRI) images. These two types of images were chosen for the amount of details that they can present using intensity values. In turn, the details that they provide lead to the use of the secondorder derivatives of the values of pixels. The second-order derivatives provide further information about the images, specifically their levels of noise and contrast. The second-order derivatives are obtained from applying a Laplacian operator on the pixels in the image. The second-order derivatives of the pixels can be collected into histograms, which are henceforth referred to as “second-order histograms”. These histograms have shape profiles with certain characteristics that are common to all images with adequate contrast and low levels of noise, e.g. profiles that approximately resemble Laplace distribution profiles. The main theory in this research is that any deviation from these presence is an indicator of the presence of factors that affect the quality of an image. One of the characteristics of the Laplace distribution profile is the convergence of two concave and non-contiguous curves into a sharp peak. The peak contains data on image details that have uniform intensity, such as regions of the same material in SEM images. Therefore, the absence of the peak in the second-order histogram of an image indicates the presence of factors that affect those details, such as noise. This is the idea behind a method to reduce noise through the restoration of the Laplace distribution profile by changing the pixel values of an image. |
|---|