The Image Retrieval Based on Scale and Rotation-Invariant Texture Features of Gabor Wavelet Transform
Autor: | Lin Xia, Chen Ning, Chen Gang |
---|---|
Rok vydání: | 2013 |
Předmět: |
business.industry
Feature extraction ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Wavelet transform Pattern recognition Gabor transform Circular shift Image texture Canberra distance Computer Science::Computer Vision and Pattern Recognition Computer vision Artificial intelligence Invariant (mathematics) business Image retrieval ComputingMethodologies_COMPUTERGRAPHICS Mathematics |
Zdroj: | 2013 Fourth World Congress on Software Engineering. |
Popis: | A scale and rotation invariant texture features extraction method is proposed and then the extracted texture features are used for image retrieval. In this method, firstly, features vector of each angle after the Gabor wavelet transform is multiplied by a Gaussian window, and then a circular shift is applied on it to shift the maximum value to be the first element, which makes the scale invariance achieved, then a circular shift is applied on the features vector to shift the maximum value to be the first element of each scale which makes the rotation invariant achieved. After that, texture features are extracted form the Gabor wavelet transform after scale and rotation invariant. Finally, the extracted texture features is used for images retrieval, the similarity is measured by Canberra distance and the retrieval effectiveness is assessed by P-R(Precision-Recall) carve and MAP(Mean Average Precision). The experimental results show that this method can accurately extract scale and rotation invariant texture features. |
Databáze: | OpenAIRE |
Externí odkaz: |