Multi-View Image Processing for 3D Shape Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual hull methods for generating three-dimensional shape data of objects with curved or concave surfaces often result in inaccuracies due to the approximation of these surfaces by flat surfaces, leading to significant errors in the generated data.
Innovation Solution
An image processing apparatus and method that utilizes multiple cameras to capture images from different viewpoints, generates approximate shape data using the visual hull method, corrects distance images based on surface three-dimensional information, and deletes unit elements not representing the object's shape using threshold values to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the visual hull method is used to generate three-dimensional shape data, then the shape can be obtained stably and fast, but the accuracy deteriorates for curved or concave surfaces due to flat surface approximation
Solution Approach 1:
The patent segments the three-dimensional shape data into multiple unit elements (voxels) and processes each element individually through evaluation and deletion operations. This segmentation allows the system to maintain the fast generation speed of the visual hull method while improving accuracy by selectively removing incorrect unit elements that cause approximation errors on curved and concave surfaces.
Solution Approach 2:
The patent implements a feedback mechanism where the generated three-dimensional shape data is evaluated based on distance images from multiple viewpoints. The evaluation results provide feedback to identify and delete unit elements that do not represent the actual object shape, thereby continuously improving measurement precision while maintaining the initial fast generation capability.
2Ease of manufacture
If the visual hull method approximates curved or concave surfaces by flat surfaces, then the processing is simple and fast, but the accuracy of the three-dimensional shape data deteriorates
Solution Approach 1:
The patent extracts and removes the incorrect unit elements (voxels) that represent the flat surface approximation errors from the three-dimensional shape data. By taking out these erroneous elements through evaluation and deletion based on multi-viewpoint distance images, the system eliminates the harmful effect of flat surface approximation while maintaining the simplicity of the initial visual hull generation process.
Solution Approach 2:
Instead of trying to improve the flat surface approximation directly, the patent inverts the approach by starting with the complete approximate shape and removing the incorrect parts. This inversion transforms the problem from one of improving approximation quality to one of eliminating errors, thereby maintaining processing simplicity while achieving higher accuracy.
3Measurement precision
If existing correction methods are used for concave portions, then some accuracy improvement is achieved, but the difference between the local shape from approximate model and original object cannot be fully made up, resulting in insufficient accuracy
Solution Approach 1:
The patent applies a universal evaluation and deletion mechanism that works for all types of surfaces including curved, concave, and complex shapes. By using distance images from multiple viewpoints and evaluating each unit element against these references, the system achieves reliable and complete shape correction across various object geometries, not just for specific cases.
Solution Approach 2:
The patent introduces the dimension of multiple viewpoints by using distance images from different camera positions. This multi-dimensional approach allows the evaluation process to detect and correct local shape differences in concave and curved portions that would be invisible from a single viewpoint, thereby achieving complete and reliable shape restoration.
Data Source
AI summary
An object is to obtain three-dimensional shape data with high accuracy from three-dimensional shape data representing an approximate shape of an object. Three-dimensional shape data of an object existing in an image capturing space is obtained. Further, a plurality of distance images each representing a distance to the object and corresponding to a plurality of viewpoints is obtained. Then, correction to evaluate the three-dimensional shape data based on the plurality of distance images and based on results of the evaluation, delete a unit element estimated not to represent a shape of the object among unit elements configuring the three-dimensional shape data is performed.