Problem
Research questions and friction points this paper is trying to address.
Automatically balances spatial and color features
Groups pixels with similar texture properties
Improves segmentation accuracy for textured images
Texture-Aware Superpixel Segmentation
🤖 AI Summary
Existing superpixel algorithms struggle to simultaneously achieve spatial coherence, color similarity, and local texture homogeneity, while heavily relying on manual parameter tuning. To address this, we propose Texture-Aware Superpixel Partitioning (TASP): (1) it adaptively models spatial constraint strength based on local feature variance, enabling texture-sensitive dynamic regularization; (2) it introduces a patch-based pixel-to-superpixel distance metric that explicitly incorporates local structural similarity; and (3) it embeds a graph-cut optimization framework to enhance segmentation robustness. TASP is the first method to jointly integrate texture homogeneity-driven distance metrics with adaptive spatial weighting into superpixel segmentation. Extensive experiments demonstrate that TASP consistently outperforms state-of-the-art methods across multiple quantitative metrics—including Boundary Recall (BE), Undersegmentation Error (UE), and Weighted Uniformity (WS)—and exhibits significant advantages on texture-rich and natural color images.
📝 Abstract
Most superpixel algorithms compute a trade-off between spatial and color features at the pixel level. Hence, they may need fine parameter tuning to balance the two measures, and highly fail to group pixels with similar local texture properties. In this paper, we address these issues with a new Texture-Aware SuperPixel (TASP) method. To accurately segment textured and smooth areas, TASP automatically adjusts its spatial constraint according to the local feature variance. Then, to ensure texture homogeneity within superpixels, a new pixel to super-pixel patch-based distance is proposed. TASP outperforms the segmentation accuracy of the state-of-the-art methods on texture and also natural color image datasets.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Texture-aware superpixel segmentation method
Automatically adjusts spatial constraint locally
Uses patch-based distance for homogeneity
🔎 Similar Papers
No similar papers found.
