Abstract:Platforms such as robots, security cameras, drones and satellites are used in multi-view imaging for three-dimensional (3D) recovery by stereoscopy or tomography. Each camera in the setup has a field of view (FOV). Multi-view analysis requires overlap of the FOVs of all cameras, or a significant subset of them. However, the success of such methods is not guaranteed, because the FOVs may not sufficiently overlap. The reason is that pointing of a camera from a mount or platform has some randomness (noise), due to imprecise platform control, typical to mechanical systems, and particularly moving systems such as satellites. So, success is probabilistic. This paper creates a framework to analyze this aspect. This is critical for setting limitations on the capabilities of imaging systems, such as resolution (pixel footprint), FOV, the size of domains that can be captured, and efficiency. The framework uses the fact that imprecise pointing can be mitigated by self-calibration - provided that there is sufficient overlap between pairs of views and sufficient visual similarity of views. We show an example considering the design of a formation of nanosatellites that seek 3D reconstruction of clouds.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that in multi - view imaging, due to the random error (noise) of camera pointing, there is insufficient field - of - view overlap, which affects the success rate of 3D reconstruction. Specifically, when platforms such as robots, security cameras, drones and satellites perform multi - view imaging, each camera has a certain field - of - view (FOV), and multi - view analysis requires sufficient overlap of these fields - of - view. However, due to the imprecise control of mechanical systems, especially for mobile platforms such as satellites, there are random errors between the actual pointing direction of the camera and the expected direction, which makes the success of multi - view imaging probabilistic. To solve this problem, this paper proposes a framework to analyze the success probability of multi - view imaging. This framework mainly focuses on the following aspects: 1. **Field - of - view overlap**: Ensure that there is a large enough overlap area between the fields - of - view of different cameras. 2. **Self - calibration**: Mitigate the impact of pointing errors through geometric self - calibration methods, provided that there is sufficient overlap between fields - of - view and sufficient visual similarity between views. 3. **Success probability**: Calculate the probability that the multi - view imaging system can successfully perform self - calibration and complete 3D reconstruction under given conditions (such as absolute pointing error APE, field - of - view size, etc.). In addition, this paper also specifically studies the application of small - satellite formations in Earth - observation missions and explores how to design these formations to improve the success rate of imaging. Through Monte Carlo simulations, the author estimates the success probability under different parameter settings and discusses the impact of these results on the design of actual missions. ### Key formulas - **Absolute - Overlap**: \[ AO=\left|\bigcap_{c = 1}^{N_{\text{cam}}}\text{FOV}[c]\right| \] - **Relative - Overlap**: \[ RO=\frac{AO}{|\text{FOV}[\text{anchor}]|} \] - **Relative pair - wise overlap between views**: \[ RO_{c,c'}=\frac{|\text{FOV}[c]\cap\text{FOV}[c']|}{\min(|\text{FOV}[c]|,|\text{FOV}[c']|)} \] - **Self - calibration success probability**: \[ P_{\text{calib}}(T||\text{FOV}[\text{anchor}]|,\text{APE},Q)\approx\frac{1}{N_{\text{MC}}}\sum_{Q' = Q}^{N_{\text{cam}}}N_B(Q') \] These formulas are used to quantify the degree of field - of - view overlap and the probability of successful self - calibration, thereby evaluating the performance of the multi - view imaging system.

Success Probability in Multi-View Imaging

Multi-view 3D Reconstruction from Uncalibrated Radially-Symmetric Cameras

Hybrid-MVS: Robust Multi-View Reconstruction with Hybrid Optimization of Visual and Depth Cues

Spatial Uncertainty Model of a Three-View RGB-D Camera System

Multi-view imaging simulation of a space born TDI CCD camera and a 3D geometric reconstruction estimate

Multi-view 3D data fusion and patching to reduce Shannon entropy in Robotic Vision

Scalable Multi-View Stereo Camera Array for Real-Time Image Capture and 3D Display in Real-World Applications

Best Viewpoint Tracking for Camera Mounted on Robotic Arm with Dynamic Obstacles

Multi-view Reconstruction under Varying Illumination Conditions

Multi-View Geometry Compression

Probabilistic Multi-View Fusion of Active Stereo Depth Maps for Robotic Bin-Picking

Prioritized multi-view stereo depth map generation using confidence prediction

3D Reconstruction through Fusion of Cross-View Images

A survey on conventional and learning‐based methods for multi‐view stereo

Multi-View Stereo: A Tutorial

A geometric approach to multi-view compressive imaging

A Probabilistic Approach for Stereo 3D Point Cloud Reconstruction from Airborne Single-Channel Multi-Aspect SAR Image Sequences

Hybrid Iteration and Optimization-based Three-dimensional Reconstruction for Space Non-Cooperative Targets with Monocular Vision and Sparse Lidar Fusion

Multi-view Monocular Depth and Uncertainty Prediction with Deep SfM in Dynamic Environments

Learning Unsupervised Multi-View Stereopsis via Robust Photometric Consistency

RIS-aided Single-frequency 3D Imaging by Exploiting Multi-view Image Correlations