3D Imager Point Cloud Generation Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing triangulation-based 3D imager devices face challenges in thermal stability, cooling efficiency, background lighting rejection, accuracy, and alignment, particularly when measuring large objects with high resolution in a short time.
Innovation Solution
A machine learning-based method is employed to generate a disparity map from pairs of images captured by a 3D imager, using a trained model that incorporates disparity neural networks or random forest algorithms to improve point cloud generation and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If traditional triangulation methods are used to measure large objects, then measurement coverage area is improved, but measurement precision deteriorates
Solution Approach 1:
The patent divides the measurement process into multiple captures with different focus settings (first focus capture and second focus capture). By segmenting the depth range into multiple focal planes and combining them, the system achieves both large measurement coverage and high precision across the entire range, resolving the contradiction between coverage area and measurement precision.
2Measurement precision
If high resolution measurement is performed on large objects, then measurement precision is improved, but measurement time increases
Solution Approach 1:
The patent performs multiple focus captures continuously without requiring physical repositioning or interruption of the measurement process. The imaging device maintains continuous operation by sequentially capturing data at different focus settings and combining them, achieving high resolution measurement of large objects while minimizing measurement time through uninterrupted data acquisition.
3Measurement precision
If multiple captures with different focus settings are performed, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent implements dynamic focus adjustment where the imaging device changes focus settings between captures. By making the focus setting a dynamic variable that can be adjusted programmatically, the system achieves enhanced measurement precision through multiple focal planes without requiring complex mechanical repositioning or additional hardware components, thus managing device complexity while improving precision.
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 enables the creation of dense, high-accuracy point clouds by enhancing the imager's ability to project and interpret light patterns, improving thermal stability, and reducing measurement time while maintaining high resolution.
Implementation Method 1
A 3D imager uses a triangulation method to measure the 3D coordinates of points on an object. The 3D imager usually includes a projector that projects onto a surface of the object either a pattern of light (also referred to as a 'light pattern') in a line or a pattern of light covering an area. A camera is coupled to the projector in a fixed relationship, for example, by attaching a camera and the projector to a common frame. The light emitted from the projector is reflected off of the object surface and detected by the camera. Since the camera and projector are arranged in a fixed relationship, the distance to the object may be determined using trigonometric principles.
Data Source
AI summary
An example method for training a machine learning model is provided. The method includes receiving training data collected by a three-dimensional (3D) imager, the training data comprising a plurality of training sets. The method further includes generating, using the training data, a machine learning model from which a disparity map can be inferred from a pair of images that capture a scene where a light pattern is projected onto an object.