PC-SAM: Patch-Constrained Fine-Grained Interactive Road Segmentation in High-Resolution Remote Sensing Images

Abstract

Road masks obtained from remote sensing images effectively support a wide range of downstream tasks. In recent years, most studies have focused on improving the performance of fully automatic segmentation models for this task, achieving significant gains. However, current fully automatic methods are still insufficient for identifying certain challenging road segments and often produce false positive and false negative regions. Moreover, fully automatic segmentation does not support local segmentation of regions of interest or refinement of existing masks. Although the SAM model is widely used as an interactive segmentation model and performs well on natural images, it shows poor performance in remote sensing road segmentation and cannot support fine-grained local refinement. To address these limitations, we propose PC-SAM, which integrates fully automatic road segmentation and interactive segmentation within a unified framework. By carefully designing a fine-tuning strategy, the influence of point prompts is constrained to their corresponding patches, overcoming the inability of the original SAM to perform fine local corrections and enabling fine-grained interactive mask refinement. Extensive experiments on several representative remote sensing road segmentation datasets demonstrate that, when combined with point prompts, PC-SAM significantly outperforms state-of-the-art fully automatic models in road mask segmentation, while also providing flexible local mask refinement and local road segmentation. The code will be available at https://github.com/Cyber-CCOrange/PC-SAM.

Figures (7)

• Figure 1: PC-SAM not only integrates fully automatic segmentation and interactive segmentation within a unified framework, but also, by explicitly constraining the refinement range of point prompts to their corresponding patches, enables local road segmentation.
• Figure 2: Overall pipeline. Stage 1 performs fully automatic road segmentation, producing both a high-recall road mask and a standard road mask. Stage 2 performs mask removal: based on the image and the fully automatic segmentation mask, users can select regions of no interest and false positive areas using negative point prompts, and the model removes the corresponding road mask regions according to these negative prompts. Stage 3 performs mask addition: based on the high-recall road mask and the mask produced in Stage 2, users can select regions of interest and false negative areas using positive point prompts, and the model segments these regions according to the positive prompts. Finally, the masks from Stage 2 and Stage 3 are fused to obtain the final segmentation result.
• Figure 3: Generation process of $M_{P}$ and $M_{N}$.
• Figure 4: Visualization of the model’s performance. In the figure, red points represent negative prompts, and green points represent positive prompts.
• Figure 5: Pipeline of morphological opening and point prompt sampling for segmentation refinement.