Intelligent Color Image Processing
Date Issued
2007
Date
2007
Author(s)
Tzeng, Mary
DOI
en-US
Abstract
With the digitization of the scientific and technological products in recent years, more and more things need the image processing skill, no matter in camera , cell-phone , projector , printer , liquid crystal TV , LCD, etc., they all need to show good-looking images or movies. In addition, the automatic image recognition is also an important subject in image processing, such as a security system that needs to recognize some biometric characteristics. However, we should analyze image and extract some characteristic features before recognition.
This thesis is composed of three parts. In first part, we discuss so-called Empirical Mode Decomposition (EMD) algorithm and apply it to analyze signals and images. EMD can decompose the signal into several Intrinsic Mode Functions (IMFs).
In second part, we discuss color transfer algorithms. Under the restriction of the environment, pictures we take are not satisfied by people due to the light source or color of scenes at that place. Color transfer algorithms not only transfer color atmosphere from image to image but also can correct scene color.
The last part discusses the issue that how to enhance images. In many situations, the camera is unable to imitate human eyes. For example, the camera can’t work well under darker environment. Here we utilize Retinex and LCRT algorithm to make pictures look clearer and keep the local contrast. In addition, when we take pictures by digital camera, the process of analog data to digital data would produce noise. Bilateral filtering can reduce the noise and keep the information of edges.
Subjects
經驗模分解
二維經驗模分解
色彩傳遞
影像加強
去雜訊
empirical mode decomposition(EMD)
bi-dimensional empirical mode decomposition (BEMD)
color transfer
retinex
local contrast range transform
bilateral filtering, image enhancement
Type
thesis