Robust Instance-Based Semi-Supervised Learning Change Detection for Remote Sensing Images
Yi Zuo,Lingling Li,Xu Liu,Zihan Gao,Licheng Jiao,Fang Liu,Shuyuan Yang
DOI: https://doi.org/10.1109/tgrs.2024.3379223
IF: 8.2
2024-03-30
IEEE Transactions on Geoscience and Remote Sensing
Abstract:Semi-supervised change detection (SSCD) has experienced rapid development, with numerous semi-supervised methods being proposed to reduce the reliance on labeled data in change detection. Existing approaches typically rely on manually set high-confidence thresholds to select robust pseudolabels. However, the single-pixel threshold filtering method for pseudolabels (STFPs) lacks context correlation, cannot eliminate high-confidence false positive (HFP) samples, and leads to erroneously filtering out low-confidence true positive (LTP) samples. To address this issue, we propose robust instance-based semi-supervised learning (RISL) change detection for remote sensing (RS) images. RISL evaluates the reliability of each instance object by linking the semantic information of the context, thereby generating robust pseudolabels. In RISL, first, a simple boundary trimming (BT) module as a preprocessing method for change prediction map is introduced. BT can effectively remove low-confidence false positive (LFP) samples while avoiding confusion in the category of instance objects, thereby improving the quality of instance objects. Then, we propose a reliable instance evaluation module (RIEM) to evaluate the reliability of each instance object. RIEM combines the semantic information of the entire instance and establishes correlations between sample contexts to determine the reliability of the instance, effectively eliminating high false positive samples. In addition, consistency regularization (CR) is integrated into RISL, and a new strategy suitable for RIEM is constructed. This strategy enhances the model's generalization ability by mining and hiding semantic information from different views of unlabeled data. Experimental results on the challenging WHU-CD, LEVIR-CD, and CDD-CD datasets show that the proposed method achieves 89.80%, 90.01%, and 87.56% scores on labeled data with 5% distribution. RISL achieves state-of-the-art performance compared to other methods.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geochemistry & geophysics