For small-satellite imagery, traditional image points derived from the direct back-projection of laser altimetry points onto the imagery often exhibit deviations of tens of pixels or more, owing to the relatively limited attitude and orbit determination accuracy of small-satellite platforms. Moreover, the diversity of imaging sensors, variations in image resolution, and inherently weak image geometric configurations further complicate the accurate acquisition of image-space coordinates for laser altimetry points. To facilitate the application of laser altimetry data for geometric positioning across multi-satellite, multi-sensor, and multi-resolution small-satellite imagery, this study proposes a measurement method for laser altimetry points tailored to small-satellite images and establishes a combined geometric positioning model that integrates virtual control points, laser altimetry points, and image-matching tie points. The framework comprises four key procedural components: (1) an image-point-measurement strategy for laser altimetry points; (2) the construction of a laser altimetry data-assisted geometric positioning model for small-satellite imagery; (3) the solution of the geometric positioning model using a total least squares approach based on the partial-EIV (errors-in-variables) models; and (4) a comprehensive accuracy assessment conducted under multiple image-combination scenarios, including single-satellite single-stereo, single-satellite multi-stereo, dual-satellite single-stereo, and multi-satellite multi-stereo imagery configurations. Experimental validation is carried out using Jilin-1 small-satellite panchromatic images (KF01A, GF02A, and GF02B) acquired over the Henan region of China. The experimental results demonstrate that, with the laser altimetry point-measurement method and the combined geometric positioning model, the vertical positioning accuracy is substantially improved across all tested image-combination scenarios. These findings further confirm the capability in enhancing the vertical geometric positioning performance of stereoscopic small-satellite imagery characterized by multi-satellite platforms, multi-sensors, and multi-resolutions over terrain conditions similar to those tested.

Existing deepfake detectors often perform well on in-domain data but generalize poorly to unseen datasets or manipulation methods. This limitation is largely attributed to their reliance on dataset-specific semantic cues rather than transferable forgery patterns. To address this limitation, we propose a generalizable deepfake detection framework that combines frequency-domain enhancement with feature disentanglement. A Phase-Amplitude Frequency Enhancement (PAFE) module enhances subtle spectral artifacts introduced during deepfake generation. We then feed the enhanced representations into an asymmetric dual-branch architecture that separates content-related information from forgery-related features. The content branch models facial semantics, while the forgery branch extracts discriminative forgery features with reduced content interference. A spatial self-attention module further refines the forgery features. We optimize the framework using image-level reconstruction loss, feature-level contrastive loss, and classification loss. Together, these objectives encourage effective feature disentanglement and improve the discriminability of the learned forgery features. Extensive experiments on several widely used deepfake benchmarks show that the proposed framework achieves competitive detection performance and improved cross-domain generalization compared with existing methods.