Telecentric Lens 3D Pose Determination Using Affine Transforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D vision systems face challenges in accurately determining the pose of objects using multiple cameras, particularly due to perspective effects and self-occlusions, which lead to decreased accuracy and increased processor overhead, especially when objects are tilted or have non-coplanar features.
Innovation Solution
The system employs non-perspective lenses, such as telecentric lenses, to acquire images that remain unchanged in size and shape regardless of the object's position, allowing for the use of affine transforms to refine feature locations across cameras, thereby improving matching accuracy and reducing search complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional perspective cameras are used to acquire images from multiple viewpoints, then 3D vision coverage is improved, but feature appearance consistency deteriorates due to perspective effects and self-occlusions
Solution Approach 1:
The patent changes the optical parameter of the camera system by using telecentric lenses instead of conventional perspective lenses. This parameter change eliminates perspective distortion and self-occlusion effects, maintaining feature appearance consistency across multiple viewpoints while still enabling 3D vision coverage through multiple camera arrangements.
2Device complexity
If the same trained model is used for all cameras in a 3D vision system, then system complexity is reduced, but searching speed and accuracy deteriorate as viewing angle increases
Solution Approach 1:
The patent makes the trained model universal across all cameras by using telecentric lenses that produce consistent feature appearances regardless of viewing angle. The same model can be applied to images from any camera in the system without degradation in searching speed or accuracy, eliminating the need for camera-specific model adaptations.
Solution Approach 2:
The patent converts the potential harm of varying viewing angles into a benefit by using telecentric lenses. Instead of requiring different models for different angles, the telecentric optics ensure that features appear consistent across all viewpoints, allowing a single model to work universally and actually improving system efficiency.
3Length of stationary object
If stereo vision systems with small baselines are used, then system compactness is improved, but measurement accuracy deteriorates to pixel level
Solution Approach 1:
The patent replaces the mechanical triangulation-based depth measurement system with an optical system using telecentric lenses. Instead of relying on small baseline triangulation which limits accuracy to pixel level, the telecentric optics provide consistent scaling and eliminate perspective distortion, enabling sub-pixel accuracy even with compact camera arrangements.
Data Source
AI summary
This invention provides a system and method for training and performing runtime 3D pose determination of an object using a plurality of camera assemblies in a 3D vision system. The cameras are arranged at different orientations with respect to a scene, so as to acquire contemporaneous images of an object, both at training and runtime. Each of the camera assemblies includes a non-perspective lens that acquires a respective non-perspective image for use in the process. The searched object features in one of the acquired non-perspective image can be used to define the expected location of object features in the second (or subsequent) non-perspective images based upon an affine transform, which is computed based upon at least a subset of the intrinsics and extrinsics of each camera. The locations of features in the second, and subsequent, non-perspective images can be refined by searching within the expected location of those images. This approach can be used in training, to generate the training model, and in runtime operating on acquired images of runtime objects. The non-perspective cameras can employ telecentric lenses.