Craniofacial reconstruction aims to estimate a person’s facial appearance from the skull and has applications in forensic investigation, reconstructive surgery, archaeology, and anthropology. The skull and face are typically represented as triangular meshes, on which geodesics provide intrinsic surface structures. Accordingly, we propose a craniofacial reconstruction framework that represents facial geometry using geodesics and learns a geodesic-based statistical model. First, we extract a set of geodesic curves for each registered face mesh, originating from the nose tip. Next, paired skull meshes and their corresponding facial geodesics are used to train a principal component analysis (PCA)-based statistical model that predicts the facial geodesics for an unseen skull. Finally, we reconstruct the full-face mesh by fitting a facial statistical shape model to the predicted geodesics. Because the intrinsic properties of the geodesics and geodesic distances are preserved under isometric deformations, the accuracy of the corresponding feature points automatically established by the geodesics and exact correspondence between the feature points are ensured, thereby improving the accuracy of the craniofacial reconstruction results. Experiments demonstrate improved reconstruction accuracy and reduced runtime compared with baseline methods.

For large-scale point cloud processing, resampling takes the important role of controlling the point number and density while keeping the geometric consistency. However, current methods cannot balance such different requirements. Particularly with large-scale point clouds, classical methods often struggle with decreased efficiency and accuracy. To address such issues, we propose a weighted Poisson-disk (WPD) resampling method to improve the usability and efficiency for the processing. We first design an initial Poisson resampling with a voxel-based estimation strategy. It is able to estimate a more accurate radius of the Poisson-disk while maintaining high efficiency. Then, we design a weighted tangent smoothing step to further optimize the Voronoi diagram for each point. At the same time, sharp features are detected and kept in the optimized results with isotropic property. Finally, we achieve a resampling copy from the original point cloud with the specified point number, uniform density, and high-quality geometric consistency. Experiments show that our method significantly improves the performance of large-scale point cloud resampling for different applications, and provides a highly practical solution.

Diffusion models have become a rising star in image super-resolution (SR) tasks. However, it is not trivial to apply diffusion models for light field (LF) image SR, which requires maintaining the high-quality visual appearance of each sub-aperture image (SAI) and the angular consistency between the different SAIs. This paper proposes the first diffusion-based LF image SR model, namely LFSRDiff, by incorporating the LF disentanglement mechanism and residual modeling. Specifically, we introduce a disentangled U-Net (Distg U-Net) for diffusion models, enabling improved extraction and fusion of the spatial and angular information in LF images. Furthermore, we leverage residual modeling in diffusion to learn the residual between the upsampled low-resolution and the ground truth high-resolution, which significantly accelerates model training and yields superior results compared to direct learning. Extensive experiments conducted on the five datasets demonstrate the effectiveness of our approach, which can produce realistic SR results and achieve the highest perceptual metric in terms of LPIPS. Code is publicly available at https://github.com/chaowentao/LFSRDiff.

Craniofacial superimposition is a crucial forensic science technique to identify human remains by matching skulls to facial images. However, this task is challenging due to significant morphological differences between skulls and faces, limited paired samples, and high data dimensionality. We proposed a geometric feature-driven metric learning method for craniofacial superimposition to address these issues. Firstly, we extracted geometric features, including depth, curvature, and elevation of 3D craniofacial data, to generate 2D maps of structured representations enriched with geometric details. Next, we novelly designed a Triplet Network for geometric feature-driven metric learning, which leverages triplet loss to learn discriminative embeddings and effectively handle the limited paired data problem. By incorporating the Sinkhorn Distance as an additional constraint, we aligned the skull and face data distributions, enhancing the matching precision. We conducted extensive experiments on a 3D craniofacial dataset, achieving a maximum accuracy of 99.45% on curvature maps, surpassing state-of-the-art methods. Our code will be available after publication at https://github.com/Lqd-js/cranial-superimposition.

Inverse rendering has been extensively explored with the advent of neural implicit fields. However, existing methods struggle to model global illumination and to fully integrate volume rendering with physically based rendering in a single stage. To address this issue, we propose TIRPL, a tailored-made inverse rendering method for point-light scenes. Our method efficiently combines volume rendering and physically based rendering to jointly optimize scene geometry, materials, and global illumination from multi-view images in a single stage. We design an indirect color network and allow the light intensity to be learnable to estimate global illumination. Additionally, a feature vector and a tailored-made light intensity decay rate are devised to combine volume rendering and physically based rendering in a single stage. Extensive experiments are conducted in both real and synthetic datasets, showing the effectiveness of our method.

3D facial reconstruction plays a crucial role in virtual reality and entertainment. Impressive rendering and animation effects have been achieved through recent advances. Existing Gaussian avatar models generate diverse expressions, but the corresponding texture changes are not so satisfactory, and their geometric structure often falls short. In response to these challenges, we propose a 3D head avatar with remarkable geometry, which can control expression variations and enable facial texture information to change along with expressions. To achieve this effect, we combine the 2D Gaussian field with the facial parametric model, use the mesh to drive Gaussian field, design a fine-tuning field for the mouth area to fit distorted expressions, and simultaneously design a color variation module to simulate changes in facial wrinkles and skin. Experiments demonstrate that our avatar model exhibits excellent performance in both appearance details and geometric shapes.

Sketch-based 3D reconstruction enables intuitive content creation through freehand drawings, yet generating high-fidelity 3D models from geometrically ambiguous, structurally simplified, and sparse sketches remain challenging. To overcome existing methods’ limitations in sketch-style generalization, contour accuracy, and suboptimal texture effects, we propose an end-to-end framework that generates textured 3D models directly from a single freehand sketch and semantic labels. To address the scarcity of paired freehand sketch training data, we introduce a 3D model-based automated sketch generation method for extracting mesh contours via a 3D mesh-to-sketch pipeline and synthesizing freehand-style sketches employing a Transformer-based stroke generator to construct a paired dataset of hand-drawn sketches and 3D models. Meanwhile, we design a contour constraint mechanism that jointly optimizes projection-space Chamfer distances and elastic metrics, significantly enhancing the reconstruction accuracy of complex geometries. Furthermore, we integrate a semantic-guided texture generation module using Text2Tex with depth-aware diffusion models and dynamic view-optimization strategies, achieving a complete geometry-appearance integrated modeling pipeline. Finally, extensive experimental results demonstrate that our method outperforms existing structural reconstruction and texture synthesis approaches, exhibiting strong generalization capabilities and practical applicability.

ABSTRACT Accurate 3D craniofacial landmark detection is critical for applications in medicine and computer animation, yet remains challenging due to the complex geometry of craniofacial structures. In this work, we propose a coarse‐to‐fine framework for anatomical landmark localization on 3D craniofacial models. First, we introduce a Diffused Two‐Stream Network (DTS‐Net) for heatmap regression, which effectively captures both local and global geometric features by integrating pointwise scalar flow, tangent space vector flow, and spectral features in the Laplace‐Beltrami space. This design enables robust representation of complex anatomical structures. Second, we propose a heat kernel‐based energy optimization method to extract landmark coordinates from the predicted heatmaps. This approach exhibits strong performance across various geometric regions, including boundaries, flat surfaces, and high‐curvature areas, ensuring accurate and consistent localization. Our method achieves state‐of‐the‐art results on both a 3D cranial dataset and the BU‐3DFE facial dataset.

As a key driving factor of land-atmosphere system, land surface temperature (LST) is widely applied in geoscience studies across various fields. Among numerous LST retrieval methods, the Split-Window (SW) algorithm has been widely used because of its advantage of free of atmospheric profile data. However, some satellites provide only one single available thermal infrared (TIR) channel, which limits the direct application of the SW algorithm. To overcome this shortcoming, this study takes Landsat-8 as an example, whose TIR channel-11 is affected by degraded calibration accuracy caused by stray light, and develops a method to construct a virtual channel using MODIS TIR data, enabling the application of the SW algorithm to Landsat-8 data for LST retrieval. During the construction, the angular normalization is adopted to the MODIS TIR data in advance. The validation results derived from the simulated dataset shows that the RMSE of LST retrieval based the virtual channel using the SW method is less than 1.2 K. Further validation with ground-based measurements from the FPK station results in an RMSE of 2.44 K, demonstrating better accuracy than the result from single channel algorithm. Moreover, the angular normalization applied to MODIS data leads to an improvement of 0.36 K in LST retrieval accuracy. The results demonstrate the advantages of LST retrieval from Landsat-8 data with virtual channel and extend the applicability of the SW algorithm.

Deducing the 3D face from a skull is a challenging task in forensic science and archaeology. This article proposes an end-to-end 3D face reconstruction pipeline and an exploration method that can conveniently create textured, realistic faces that match the given skull. To this end, we propose a tissue-guided face creation and adaptation scheme. With the help of the state-of-the-art text-to-image diffusion model and parametric face model, we first generate an initial reference 3D face, whose biological profile aligns with the given skull. Then, with the help of tissue thickness distribution, we modify these initial faces to match the skull through a latent optimization process. The joint distribution of tissue thickness is learned on a set of skull landmarks using a collection of scanned skull-face pairs. We also develop an efficient face adaptation tool to allow users to interactively adjust tissue thickness either globally or at local regions to explore different plausible faces. Experiments conducted on a real skull-face dataset demonstrated the effectiveness of our proposed pipeline in terms of reconstruction accuracy, diversity, and stability.

3D face shape completion is a necessary pre-process for various facial applications as they are often mutilated due to the acquiring environment or occlusion. However, it is challenging to ensure identity consistency when completing faces with large missing regions. Therefore, we introduce craniofacial information to supervise the completion of face shapes. Firstly, a novel dual encoder-decoder structure for face depth image inpainting is constructed by combining dilated convolution and the coherent semantic attention mechanism, guaranteed to generate smooth results with complete semantic information even in the presence of large missing regions in the face model. Then, we innovatively design a craniofacial superimposition module to determine the probability that the inpainted face and corresponding skull come from the same person, constraining the inpainting network to learn identity consistency information. Finally, extensive experimental results show that our method can effectively complete 3D face shapes containing large arbitrary missing regions while guaranteeing identity consistency.

Point cloud completion aims at recovering the complete point cloud from an incomplete input. A general scheme is to generate a group of coarse points first that generalize the global shape, and then reconstruct dense point cloud by upsampling operation. In this paper, we propose a novel point cloud completion network, SkeletonFormer, to tackle two critical challenges: fully utilizing the information from the point cloud with various incompleteness degree and recovering high-quality geometric structures. To increase the universality of our model to diverse input, we propose a score mechanism to dynamically select proper skeleton points that can adapt to various degree of deficiency. To improve the perception of the target object, we use self-projected depth images as an augmented modality representation to observe the input. A modality unification module is designed to fuse the depth image feature and the point cloud feature, and it can alleviate the intrinsic differences among multi-modal information. The fused feature is used to assist the prediction of the skeleton points that represent the holistic complete object. Furthermore, by fully leveraging local geometric information, we design a novel and effective DeconvNet to reconstruct fine-grained patterns around the skeleton points. Extensive experiments demonstrate that our SkeletonFormer surpasses existing works by a large margin and achieves state-of-the-art performance on various benchmarks.

Facial expressions are an essential part of human emotional communication, and micro-expressions (MEs), as transient and imperceptible non-verbal signals, can potentially reveal real human emotions. However, subtle motion variations, limited and unbalanced samples make micro-expression recognition (MER) challenging. In this paper, we design a novel dual-branch learning framework of multi-level flow-driven attention for micro-expression recognition (MFDAN), which innovatively integrates optical flow prior to guide the attention learning in the image encoding branch, enabling the model to focus on the most discriminative facial regions for subtle motion patterns. Firstly, we extract optical flow information by an optical flow encoding module. Then, in the image coding module, we construct a Transformer structure containing an optical flow-driven attention mechanism, which can effectively locate the interest region of micro-expressions in the image according to the position information of optical flow to capture more sensitive and fine-grained micro-expressions. By interoperating prior knowledge with data learning, and introducing the Dropkey operation and Focal Loss, our method can handle subtle micro-expression features on small imbalanced datasets. Through extensive experiments on three independent datasets and a composite database, including SMIC-HS, SAMM, and CASME II, robust leave-one-subject-out (LOSO) evaluation results show that our method outperforms state-of-the-art methods especially on the composite database.

This article proposes a combination method of nonlinear split-window (NSW) algorithm and temperature and emissivity separation (TES) algorithm to estimate land surface temperature (LST) from the remotely sensed data observed by the Advanced Geosynchronous Radiation Imager (AGRI) onboard Fengyun-4B (FY-4B), China’s second-generation meteorological geostationary satellite. The atmospheric radiation transfer model MODTRAN5.2 is used to simulate the AGRI thermal infrared (TIR) channel satellite observations in ten different viewing zenith angles (VZAs) from 0° to 70°. The optimal thermal channel combination and coefficients of the NSW algorithm are determined using a statistical regression method according to the grouping of the mean emissivity, the atmospheric water vapor content (WVC), and the LST. ERA5 reanalysis data provide atmospheric profiles for atmospheric correction, and then, the land surface emissivity (LSE) could be estimated according to the TES algorithm. The combination of Channel-12 (centered at$8.55~\mu \text{m}$) and Channel-14 (centered at$12.00~\mu \text{m}$) or the combination of Channel-13 (centered at$10.80~\mu \text{m}$) and Channel-14 (centered at$12.00~\mu \text{m}$) depends on different groups and VZAs. The statistical regression analysis showed that the root-mean-square error (RMSE) between the simulated and estimated LST is less than 0.7 and 1.8 K with the determined emissivity under VZA = 0° and VZA = 60°, respectively. Compared with the MODIS LST products (MYD11A1), the retrieved LST image has a similar spatial distribution, with the RMSE of 1.71 K.

As a significant geometric feature of 3D point clouds, sharp features play an important role in shape analysis, 3D reconstruction, registration, localization, etc. Current sharp feature detection methods are still sensitive to the quality of the input point cloud, and the detection performance is affected by random noisy points and non-uniform densities. In this paper, using the prior knowledge of geometric features, we propose a Multi-scale Laplace Network (MSL-Net), a new deep-learning-based method based on an intrinsic neighbor shape descriptor, to detect sharp features from 3D point clouds. First, we establish a discrete intrinsic neighborhood of the point cloud based on the Laplacian graph, which reduces the error of local implicit surface estimation. Then, we design a new intrinsic shape descriptor based on the intrinsic neighborhood, combined with enhanced normal extraction and cosine-based field estimation function. Finally, we present the backbone of MSL-Net based on the intrinsic shape descriptor. Benefiting from the intrinsic neighborhood and shape descriptor, our MSL-Net has simple architecture and is capable of establishing accurate feature prediction that satisfies the manifold distribution while avoiding complex intrinsic metric calculations. Extensive experimental results demonstrate that with the multi-scale structure, MSL-Net has a strong analytical ability for local perturbations of point clouds. Compared with state-of-the-art methods, our MSL-Net is more robust and accurate.

Landmark localization plays a significant role in craniofacial registration, reconstruction, and authentication. The key challenges for localizing landmarks on point cloud craniofacial models include irregular structures, non-uniform densities, and uncertain local regions. In this paper, we propose an end-to-end regression network that can directly estimate craniofacial landmarks on point cloud models. The proposed network utilizes edge convolution to extract local features and pooling layers to aggregate global features. It realizes the end-to-end regression for landmark localization. Experimental results demonstrate that our method is robust on point clouds with sparse and unevenly distributed sampling. It can produce accurate, controllable, and efficient 3D landmarks.

Craniofacial reconstruction is fundamental in resolving forensic cases. It is rather challenging due to the complex topology of the craniofacial model and the ambiguous relationship between a skull and the corresponding face. In this paper, we propose a novel approach for 3D craniofacial reconstruction by utilizing Conditional Generative Adversarial Networks (CGAN) based on craniofacial depth map. More specifically, we treat craniofacial reconstruction as a mapping problem from skull to face. We represent 3D cran- iofacial shapes with depth maps, which include most craniofacial features for identification purposes and are easy to generate and apply to neural networks. We designed an end-to-end neural networks model based on CGAN then trained the model with paired craniofacial data to automatically learn the complex nonlinear relationship between skull and face. By introducing body mass index classes(BMIC) into CGAN, we can realize objective reconstruction of 3D facial geometry according to its skull, which is a complicated 3D shape generation task with different topologies. Through comparative experiments, our method shows accuracy and verisimilitude in craniofacial reconstruction results.

Skull registration plays a fundamental role in forensic science and is crucial for craniofacial reconstruction. The complicated topology, lack of anatomical features, and low quality reconstructed mesh make skull registration challenging. In this work, we propose an automatic skull registration method based on the discrete uniformization theory, which can handle complicated topologies and is robust to low quality meshes. We apply dynamic Yamabe flow to realize discrete uniformization, which modifies the mesh combinatorial structure during the flow and conformally maps the multiply connected skull surface onto a planar disk with circular holes. The 3D surfaces can be registered by matching their planar images using harmonic maps. This method is rigorous with theoretic guarantee, automatic without user intervention, and robust to low mesh quality. Our experimental results demonstrate the efficiency and efficacy of the method.

Non-rigid 3D shape correspondence is a fundamental and challenging problem. Isometric correspondence is an important topic because of its wide applications. But it is a NP hard problem if you detect the mapping directly. In this paper, we propose a novel approach to find the correspondence between two (nearly) isometric shapes. Our method is based on the geodesic structure of the shape and minimum cost flow. Firstly, several pre-computed base vertices are initialized for embedding the shapes into Euclidian space, which is constructed by the geodesic distances. Then we construct a network flow with the points of the two shapes and another two virtual points, source point and sink point. The arcs of the network flow are the edges between each point on two shapes. And the L2distances in the k dimensional Euclidian embedding space are taken as the arc costs and a capacity value is added on each point in the above network flow. At last we solve the correspondence problem as a minimum cost max flow problem (MCFP) with shortest path faster algorithm (SPFA). Experiments show that our method is accurate and efficient.

In shape analysis, scaling factors have a great influence on the results of non-rigid shape retrieval and comparison. In order to eliminate the scale ambiguity in shape acquisition and other cases, a method with scale-invariant property is required for shape analysis. The mapping method previously proposed only preserves geodesic distances between pair wise points. In this paper, a Scale-invariant Heat Kernel Mapping (SIHKM) method is introduced, which bases on the Scale-invariant Heat Kernel (SIHK) that handles various types of 3D shapes with different kinds of scaling transformations. SIHK is the generalization of the Heat Kernel and related to the heat diffusion behavior on shape. By using the SIHK, we retrieve intrinsic information from the scaled shapes while ignoring the impact of their scaling. SIHKM method maintains the heat kernel between two corresponding points on the shape with scaling deformations, including scaling transformation only, isometric deformation and scaling, and local scaling on shapes. The proof of the theory and experiments are given in this work. The experiments are performed on the TOSCA dataset, which show that our proposed method achieves good robustness and effectiveness to scaled shape analysis.