Content based logo image retrieval using shape and color

Pao Yue-kong Library Electronic Theses Database

Content based logo image retrieval using shape and color


Author: Chan, Pui-shan Peggy
Title: Content based logo image retrieval using shape and color
Degree: M.Sc.
Year: 2001
Subject: Hong Kong Polytechnic University -- Dissertations
Image transmission
Image processing -- Digital techniques
Image reconstruction
Department: Multi-disciplinary Studies
Dept. of Computing
Pages: viii, 111, 3 leaves : ill. (some col.) ; 30 cm
Language: English
InnoPac Record:
Abstract: Finding similar shapes in image databases is a challenging topic in content-based retrieval. This project deals with efficient retrieval of images from logo image databases based on the color, shape feature and Zernike moments in images. Unlike most solutions to this problem, an attempt has been made to integrate the image representation by using color, shape and Zernike moments as principle attributes for images feature extraction so as to cope with changes in rotation, scale and translation. An ideal content-based retrieval system requires robustness, efficiency, accuracy and tolerance to noise. The recognition of an image independent of its position, size, scale, orientation is important in the field of pattern recognition. A number of techniques have been developed to extract features invariant with respect to image translation, change of scale and rotation. A single attribute is not enough to represent detail information of an image like color information and spatial information. Thus, this cannot fulfill all the requirements of an ideal content-based retrieval system such that images can be accurately and efficiently retrieved from images that are shifted, rotated, magnified, distorted, or even contaminated with noise. Since color plays an important role in feature attributes describing an image, it must be taken into account to represent images. Due to the drawback of the rotation invariance of color histograms, fuzzy approach is introduced in image query by color. The image will be further divided into nine sections. Each section again represented by corresponding color histograms so that rotation images can be filtered out by matching color histograms of the whole images and those color histograms for the corresponding nine partial sections of the images. Pinpointing to the insufficiency of edge direction histogram as shape attributes representing an image, Zernike moments are used to capture more spatial information of the image. The reason to use Zernike moments and not other complex valued moments is that the Zernike moments are orthogonal on the unit disk. It is thus straightforward to reconstruct an image function given the Zernike moments up to any order. Incorporating rotation invariance in shape matching generally increases the computation requirement. The magnitude of the Zernike moments has been successfully used as rotation invariant feature for object recognition. The phase angles of the Zernike moments are used to determine the orientation of the identified object. Saving of computational time of Zernike moments can be achieved by using the explicit forms of their radial polynomials. The performance of the system in term of precision and recall rate is estimated. Query by both color and shape performs very well in similar image retrieval. It is concluded that Zernike moments are the most accurate query parameter in retrieving from noisy images. The phase differences of the argument of the Zernike moments among the images of the same magnitude of Zernike moments give information in how many degrees of rotation that those images have compared with the query image. However, query by color with multi-color histograms cannot discriminate rotated images from the similar ones retrieved.

Files in this item

Files Size Format
b16681514.pdf 12.84Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.


Quick Search


More Information