Wave Field Reconstruction via Optical Perception and Depth Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wave field reconstruction technologies face challenges in achieving accurate and real-time reconstruction of water surface wave fields, particularly in outdoor environments, due to limitations in perception and high computational requirements.
Innovation Solution
A wave field reconstruction method based on optical perception is developed, which includes building a data generation platform for a virtual wave field, constructing domain converter and depth estimation modules, and using a point cloud mapping algorithm to generate surface point cloud data from real water surface images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If numerical calculation is carried out in combination with physical equations during image capture, then wave field reconstruction accuracy is improved, but calculation time increases and real-time reconstruction cannot be achieved
Solution Approach 1:
The patent pre-calculates and stores wave field data in a lookup table during an offline phase. During real-time operation, the system only needs to query pre-computed values based on captured images, avoiding time-consuming numerical calculations while maintaining reconstruction accuracy through the pre-stored relationship between image features and wave field parameters.
Solution Approach 2:
The patent replaces the traditional numerical calculation approach (mechanical/computational system) with a data-driven query approach using pre-computed lookup tables. This substitution transforms the real-time reconstruction process from intensive computation to efficient data retrieval, achieving both accuracy and speed requirements.
2Device complexity
If SFS algorithm is used for wave field reconstruction, then computational complexity is reduced, but projection occlusion problems cannot be solved
Solution Approach 1:
The patent introduces a depth estimation module as an intermediary component between image capture and wave field reconstruction. This module first estimates depth information from captured images, then uses the depth data to guide the reconstruction process, effectively handling projection occlusion issues while keeping the overall system complexity manageable through modular design.
Solution Approach 2:
The patent divides the wave field reconstruction process into separate functional modules: image capture, depth estimation, and wave field reconstruction. This segmentation allows each module to be optimized independently, with the depth estimation module specifically addressing occlusion problems without increasing the complexity of the reconstruction algorithm itself.
3Device complexity
If traditional perception methods are used for unmanned boats, then system simplicity is maintained, but movement accuracy control is limited
Solution Approach 1:
The patent makes the perception system multi-functional by integrating wave field reconstruction capabilities that serve both navigation and environmental understanding purposes. The same image capture and depth estimation modules used for wave field analysis also provide spatial awareness for movement control, eliminating the need for separate specialized sensors and maintaining system simplicity while enhancing accuracy.
Data Source
AI summary
This invention discloses a wave field reconstruction method via optical perception, including: (1) building a data platform for a virtual wave field; (2) pre-training a domain converter module using color image data of the virtual wave field and real water surface image data captured by a camera; (3) pre-training a depth estimation module using generated paired color image data and depth image data of the virtual wave field; (4) converting the style of a real water surface image into a virtual-water-surface-style image via the domain converter module; (5) outputting the virtual-water-surface-style image as a depth image with the same style and recording the distance between a wave surface sampling point and the camera's optical center; and (6) generating surface point cloud data of a water surface wave field with the camera's optical center as the coordinate origin. The method enables real-time, robust wave field depth reconstruction, ensuring reliability and authenticity.