Texture Feature Representation based on Gabor Wavelet Decomposition

People

Wei Ma, B.S.Manjunath

Objective

We have proposed the use of Gabor wavelet features for texture analysis and provide a comprehensive experimental evaluation of this algorithm on a large texture image database.

The basic idea is to extract features at multiple scales and orientations using a Gabor wavelet decomposition. These features compare favorably with other existing texture analysis algorithms. In particular, comparisons are made with orthogonal wavelet transform features and multi-resolution simultaneous autoregressive model features. Our experimental results indicate that the Gabor features have the best pattern retrieval performance.

Software

» Feature Extraction and Gabor Filtering

Datasets

» Texture features of the Brodatz album (1856 feature vectors (rows) - each containing 48 dim vector)

Explanation

There are six different texture feature extraction schemes included in the evaluation:

  1. Gabor wavelet transform features
  2. Multi-resolution simultaneous autoregressive model features
  3. Orthogonal wavelet transform features
  4. Bi-orthogonal wavelet transform features
  5. Tree-structured decomposition using orthogonal filter bank
  6. Tree-structured decomposition using bi-orthogonal filter bank

The texture database used in the experiments consists of 116 different large texture images. Each one is 512x512 pixels and is divided into 16 non-overlapping subimages, each 128x128 pixels in size, thus creating a database of 1856 texture images. The following figure shows the demo program which is used to compare the performance of different texture features. The upper half part of demo shows the samples from the 116 texture classes, and the lower half part displays the query pattern (upper left) and its top matches from the database. They are shown in row scan order with increasing distance.

Pattern Retrievals using Gabor Features

Example 1:

Example 2:

Example 3:

Publications

These materials are presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each authors copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder. Wei-Ying Ma, Manjunath, B.S. " A texture thesaurus for browsing large aerial photographs ", Journal of the American Society for Information Science, vol.49, (no.7), Wiley for ASIS, pp.633-48, May 1998. [abstract] B. S. Manjunath and W.Y. Ma, "Texture features for browsing and retrieval of image data", IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), vol.18, no.8, pp.837-42, Aug 1996. (corrections and derivation of Gabor filter dictionary parameters) [abstract] W. Y. Ma and B. S. Manjunath, "Texture features and learning similarity," Proc. IEEE International Conference on Computer Vision and Pattern Recognition, San Francisco, CA, pp. 425-430, June 1996. [abstract]

See also