Neural See-Through Sensing for Layered 3D Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
See-through sensing technologies, such as THz sensing, face challenges in reconstructing the three-dimensional structure of target objects due to complexity in image reconstruction and undesirable artifacts, particularly exacerbated by the shadow effect caused by non-uniform wave penetration through layered structures, which degrades image quality, especially in deeper layers.
Innovation Solution
A system and method that treat the target object as a multi-layer structure, using a neural network to classify waves modified by penetration, allowing for parallel wave emission and measurement to reconstruct images of each layer individually or jointly, leveraging the unique and stable modification of waves by different layers to improve image quality and accuracy, and employing techniques like pre-training and combinatory sections to enhance classification and reduce artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the target object is treated as a multi-layer structure for image reconstruction, then computational performance is improved, but image quality of deeper layers deteriorates due to shadow effect
Solution Approach 1:
The patent divides the three-dimensional object into multiple two-dimensional layers along the wave propagation direction. Each layer is reconstructed independently or jointly, transforming a complex 3D reconstruction problem into simpler 2D problems. This segmentation improves computational efficiency while the patent addresses the shadow effect through specialized processing of deeper layers
Solution Approach 2:
The patent introduces an intermediary processing step that accounts for the shadow effect caused by upper layers. By modeling and compensating for the non-uniform wave penetration through intermediate layers, the system can reconstruct deeper layers with improved accuracy, effectively using the intermediary shadow effect as information rather than treating it as pure noise
2Measurement precision
If hierarchical image recovery techniques are used to increase image quality, then reconstruction quality improves, but computational complexity increases
Solution Approach 1:
The patent segments the reconstruction problem into layer-specific subproblems, where each layer can be processed with appropriate complexity. This allows simpler processing for layers that don't require high fidelity and more sophisticated processing only where needed, balancing overall computational complexity with reconstruction quality
Solution Approach 2:
The patent applies different processing strategies to different layers based on their specific characteristics and importance. Deeper layers that suffer from shadow effects may receive more sophisticated processing, while superficial layers can use simpler methods, optimizing the trade-off between reconstruction quality and computational complexity locally rather than uniformly across all layers
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves computational performance and image quality by addressing the shadow effect and enhancing sparsity in image reconstruction, allowing for more accurate classification and reconstruction of layered structures, even in the presence of non-uniform penetration, thereby improving the overall quality of 3D image reconstruction.
Implementation Method 1
an emitter configured to emit a set of waves in parallel directions of propagation to penetrate a sequence of layers of the target object
Implementation Method 2
a receiver configured to measure intensities of the set of waves modified by penetration through the layers of the target object
Data Source
AI summary
A scanner for image reconstruction of a structure of a target object uses a neural network trained to classify each segment of a sequence of segments of a modified wave into one or multiple classes. The sequence of segments corresponds to the sequence of layers of the target object, such that a segment of modified wave corresponds to a layer having the same index in the sequence of layers as an index of the segment in the sequence of segments. The scanner executes the neural network for each wave modified by penetration through the layers of the target object to produce the classes of segments of the modified waves. Next, the scanner selects the classes of segments of different modified waves corresponding to the same layer to produce an image of the layer of the target object with pixel values being functions of labels of the selected classes.