Class-Aware Semantic Calibration for Cross-Scene Hyperspectral Image Classification.
Saved in:
| Title: | Class-Aware Semantic Calibration for Cross-Scene Hyperspectral Image Classification. |
|---|---|
| Authors: | Shi, Boshan1 (AUTHOR), Liu, Yanbo2 (AUTHOR), Zhang, Youqiang3 (AUTHOR), Cao, Guo1 (AUTHOR) caoguo@njust.edu.cn |
| Source: | Remote Sensing. Jun2026, Vol. 18 Issue 12, p1976. 26p. |
| Subjects: | Remote sensing, Image recognition (Computer vision) |
| Abstract: | Highlights: What are the main findings? A class-aware semantic calibration framework, CASC-DA, is proposed for single-source cross-scene hyperspectral image classification. CASC-DA calibrates three observable prediction structure distortions: source prior bias, boundary ambiguity, and prediction-level dependency drift. Experiments on three cross-scene HSI benchmarks show that CASC-DA consistently improves generalization over representative domain generalization methods. What are the implications of the main findings? Semantic structure calibration is complementary to pseudo-source augmentation and can further improve cross-scene robustness. The proposed calibration and alignment terms are lightweight training objectives and do not introduce additional inference time complexity. The results suggest that robust HSI domain generalization should consider not only feature invariance but also class-level and prediction-level structural reliability. Cross-scene Hyperspectral Image (HSI) classification faces substantial domain shifts caused by sensor heterogeneity, acquisition variation, and scene diversity. While benchmark annotations are assigned to individual center pixels, local patches often contain implicit multi-label semantics due to spectral mixing and spatial overlap. This mismatch distorts prediction structure, exacerbates generalization errors, and limits the effectiveness of standard domain generalization (DG) techniques focused solely on feature or prediction invariance. We propose Class-Aware Semantic Calibration (CASC), a systematic semantic structure calibration framework that addresses three complementary distortions induced by mismatched patch supervision: (i) Balance corrects class frequency bias via reweighted supervision; (ii) Separability enhances boundary decision stability through margin-based logit calibration; and (iii) Independence reduces domain-specific spurious co-occurrence via prediction covariance decorrelation. To preserve calibrated semantics under pseudo-source shift, we further introduce a complementary DualAlign (DA) module, which jointly aligns feature statistics and prediction distributions, enforcing consistency at both representation and semantic levels. Extensive experiments on three cross-scene benchmarks (Houston, Pavia, and WHU-Hi) demonstrate that CASC-DA consistently improves performance over strong baselines, achieving an average gain of 3.0% in overall accuracy and 4.9% in Kappa coefficient compared with the best-performing baseline on each dataset. These results underscore the importance of semantic structure calibration for domain-generalized HSI classification. [ABSTRACT FROM AUTHOR] |
| Copyright of Remote Sensing is the property of MDPI and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | Highlights: What are the main findings? A class-aware semantic calibration framework, CASC-DA, is proposed for single-source cross-scene hyperspectral image classification. CASC-DA calibrates three observable prediction structure distortions: source prior bias, boundary ambiguity, and prediction-level dependency drift. Experiments on three cross-scene HSI benchmarks show that CASC-DA consistently improves generalization over representative domain generalization methods. What are the implications of the main findings? Semantic structure calibration is complementary to pseudo-source augmentation and can further improve cross-scene robustness. The proposed calibration and alignment terms are lightweight training objectives and do not introduce additional inference time complexity. The results suggest that robust HSI domain generalization should consider not only feature invariance but also class-level and prediction-level structural reliability. Cross-scene Hyperspectral Image (HSI) classification faces substantial domain shifts caused by sensor heterogeneity, acquisition variation, and scene diversity. While benchmark annotations are assigned to individual center pixels, local patches often contain implicit multi-label semantics due to spectral mixing and spatial overlap. This mismatch distorts prediction structure, exacerbates generalization errors, and limits the effectiveness of standard domain generalization (DG) techniques focused solely on feature or prediction invariance. We propose Class-Aware Semantic Calibration (CASC), a systematic semantic structure calibration framework that addresses three complementary distortions induced by mismatched patch supervision: (i) Balance corrects class frequency bias via reweighted supervision; (ii) Separability enhances boundary decision stability through margin-based logit calibration; and (iii) Independence reduces domain-specific spurious co-occurrence via prediction covariance decorrelation. To preserve calibrated semantics under pseudo-source shift, we further introduce a complementary DualAlign (DA) module, which jointly aligns feature statistics and prediction distributions, enforcing consistency at both representation and semantic levels. Extensive experiments on three cross-scene benchmarks (Houston, Pavia, and WHU-Hi) demonstrate that CASC-DA consistently improves performance over strong baselines, achieving an average gain of 3.0% in overall accuracy and 4.9% in Kappa coefficient compared with the best-performing baseline on each dataset. These results underscore the importance of semantic structure calibration for domain-generalized HSI classification. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 20724292 |
| DOI: | 10.3390/rs18121976 |
