Texture classification using block intensity and gradient difference (BIGD) descriptor
作者：
Highlights：
• We present a local descriptor called block intensity and gradient difference BIGD.
• We sample multi-scale block pairs and utilize intensity and gradient differences.
• BIGD captures patch patterns at different orientations and spatial granularities.
• We evaluate BIGD and other peer methods on texture datasets through classification.
• BIGD obtains 0.12%∼6.43% higher accuracy than DMD showing its discriminative ability.
摘要
•We present a local descriptor called block intensity and gradient difference BIGD.•We sample multi-scale block pairs and utilize intensity and gradient differences.•BIGD captures patch patterns at different orientations and spatial granularities.•We evaluate BIGD and other peer methods on texture datasets through classification.•BIGD obtains 0.12%∼6.43% higher accuracy than DMD showing its discriminative ability.
论文关键词：Local descriptor,Block intensity and gradient difference (BIGD),Local feature extraction,Multi-scale,Texture classification
论文评审过程：Received 8 April 2019, Revised 28 December 2019, Accepted 28 December 2019, Available online 3 January 2020, Version of Record 25 January 2020.
论文官网地址：https://doi.org/10.1016/j.image.2019.115770
