Bilateral Local Binary Patterns for Rotation Invariant Texture Retrieval
| Autor: | Mingang Zhou, Zhang Jiuwen, Runpu Zhang |
|---|---|
| Rok vydání: | 2015 |
| Předmět: |
Pixel
business.industry Local binary patterns Feature vector ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Pattern recognition Wavelet Image texture Computer Science::Computer Vision and Pattern Recognition Histogram Computer vision Artificial intelligence Invariant (mathematics) business Image retrieval Mathematics |
| Zdroj: | 2015 2nd International Conference on Information Science and Control Engineering. |
| DOI: | 10.1109/icisce.2015.44 |
| Popis: | In this paper, a novel local descriptor, named bilateral local binary patterns (BLBP) is proposed for rotation invariant texture retrieval. The proposed BLBP is based on the Local Binary Pattern operator (LBP) of combining phase and module information of images. For phase information, we use a projecting filter firstly to obtain an analytic signal of the input image and so the obtained image has explicit phase information, then the LBP operator is applied on it in order to achieve a histogram of phase to describe the local pattern of phase. For module information, we use LBP operator directly, then we can achieve a histogram of module to describe the local pattern of pixel variation. We introduce a linear relation between the two histograms to achieve a new histogram as the feature vector of texture image. Experimental results obtained from two databases demonstrate that the proposed BLBP can achieve higher retrieval rate than only using LBP for original images while the computational complexity is lower than using the phase information with LBP in the complex wavelet domain for rotation invariant image retrieval. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|