Adopting Hybrid Descriptors to Recognise Leaf Images for Automatic Plant Specie Identification
| Autor: | Ali A. Al-Kharaz, Raid Lafta, Ji Zhang, Xiaohui Tao |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Pixel
Contextual image classification business.industry Local binary patterns Computer science ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 02 engineering and technology HSL and HSV Image (mathematics) Identification (information) Wavelet 020204 information systems Histogram 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Computer vision Artificial intelligence business |
| Zdroj: | Advanced Data Mining and Applications ISBN: 9783319495859 ADMA |
| Popis: | In recent years, leaf image recognition and classification has become one of the most important subjects in computer vision. Many approaches have been proposed to recognise and classify leaf images relying on features extraction and selection algorithms. In this paper, a concept of distinctive hybrid descriptor is proposed consisting of both global and local features. HSV Colour histogram (HSV-CH) is extracted from leaf images as the global features, whereas Local Binary Pattern after two level wavelet decomposition (WavLBP) is extracted to represent the local characteristics of leaf images. A hybrid method, namely “Hybrid Descriptor” (HD), is then proposed considering both the global and local features. The proposed method has been empirically evaluated using four data sets of leaf images with 256 \(\times \) 256 pixels. Experimental results indicate that the performance of proposed method is promising – the HD outperformed typical leaf image recognising approaches as baseline models in experiments. The presented work makes clear, significant contribution to knowledge advancement in leaf recognition and image classification. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|