Optimization of Image Zernike Moments Shape Feature Based on Evolutionary Computation

Optimization of Image Zernike Moments Shape Feature Based on Evolutionary Computation

Maofu Liu (Wuhan University of Science and Technology, China) and Huijun Hu (Wuhan University of Science and Technology, China)
DOI: 10.4018/978-1-4666-3958-4.ch008
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

The image shape feature can be described by the image Zernike moments. In this chapter, the authors point out the problem that the high dimension image Zernike moments shape feature vector can describe more detail of the original image but has too many elements making trouble for the next image analysis phases. Then the low dimension image Zernike moments shape feature vector should be improved and optimized to describe more detail of the original image. Therefore, the optimization algorithm based on evolutionary computation is designed and implemented in this chapter to solve this problem. The experimental results demonstrate the feasibility of the optimization algorithm.
Chapter Preview
Top

Introduction

With the development and maturity of the image acquisition and storage technology, especially the sharp emergency and increment of Web images in the social networking service environment, Web users and humans are deluged by a great quantity of image data. They need the techniques and tools to efficiently retrieve, analyze, and understand the image data. Then the image feature which can be used to represent the image becomes more and more important. In fact, the image feature extraction and description is the most critical phase in the image retrieval, image analysis, and image understanding.

The image features include text, color, texture, shape, edge, shadows, temporal detail, and so on (Foschi, Kolippakkam, Huan, et al., 2002) and the image shape feature plays a very fundamental and important role in image analysis and understanding, so an effective and efficient shape descriptor is the key component of the image shape feature extraction and representation.

There are two types of image shape feature descriptors: contour-based and region-based shape feature descriptors. The region-based shape feature descriptors, for example the moment, are more reliable for shapes that have complex boundaries, because they rely on not only the contour pixels but also all pixels constituting the shapes (Teh & Chin, 1988).

The moments, especially geometric moment, centric moment, orthogonal invariance moments, have already been used in image shape description and content-based image retrieval (Teh & Chin, 1988; Liao & Pawlak, 1996; Mukundan & Pang, 2002; Pang, Andrew, David, et al., 2004). The Zernike moment, one kind of the orthogonal invariance moments, is the most commonly used technique in image shape feature extraction and description. Many researchers have paid much more attention to its invariant characteristics, including translation invariance, rotation invariance and scale invariance (Liao & Pawlak, 1996, 1997; Kim & Kim, 2000; Palak & Subbalakshmi, 2004; Chalechale, Naghdy, & Mertins, 2005; Kamila, Mahapatra, & Nanda, 2005).

After the computation of the Zernike moments to the original image, the image shape feature vector can be obtained. In fact, much higher order of the Zernike moments are used, more detail of the original image shape can be described. On the other hand, given the image Zernike moments shape feature vector and using the image reconstruction technique, the reconstructed image based on the Zernike moments shape feature vector of the original image can be turned out.

The reconstructed image will contain more detail of the original image and be closer to the original image as the order is higher, but the dimension of the Zernike moments shape feature vector also will be higher. If the order of the Zernike moment is 20, 25, 30, 35, 40, and 45, the dimension of the Zernike moments shape feature vector will correspondingly be 121, 182, 256, 342, 441, and 552.

If wanting to describe more detail of the original image shape, higher order of the Zernike moment will be used and higher dimension of the image Zernike moments shape feature vector will be obtained. The high dimension of the feature vector will make trouble to the next phases of image analysis, so the researchers have to solve the problem that the more detail can be described via the lower order of the Zernike moment and lower dimension of image Zernike moments shape feature vector.

In the chapter, the optimization algorithm based on evolutionary computation is designed and the low dimension image Zernike moments shape feature vector will be improved and optimized to describe more detail of the original image.

Complete Chapter List

Search this Book:
Reset