Texture classification by multi-predicate local binary pattern operators

Key Points

• This research aims to enhance texture classification in images by introducing a multi-predicate local binary pattern approach.
• Benchmarking the original local binary pattern approach against the multi-predicate version.
• Focusing on supervised texture segmentation problems as outlined in a prior comparative study.
• Analyzing performance on images with textures across multiple scales.
• The multi-predicate LBP outperforms the simple LBP in segmenting textures.
• Improved accuracy in identifying textures at different scales, enhancing overall image processing capabilities.