Object Recognition Using Depth Image Hole Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for recognizing objects using depth images face errors due to missing data, particularly in applications like video games and computer graphics, where accurate motion estimation in 3D spaces is crucial but hindered by holes in depth images.
Innovation Solution
An apparatus and method that include a foreground extractor, a hole determiner, a feature vector generator, and an object recognizer, which utilize both depth and color images to identify and account for holes in depth images by generating feature vectors and comparing them to reference vectors, employing decision trees and hierarchical clustering to improve object recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If depth image is used for object recognition, then 3D motion estimation capability is improved, but recognition accuracy deteriorates due to missing data (holes) in depth image
Solution Approach 1:
The patent combines depth image and color image to compensate for the limitations of using only depth image. By merging information from both images, the system achieves accurate object recognition even when depth data has holes or missing values, thus resolving the contradiction between 3D capability and recognition accuracy
Solution Approach 2:
The patent introduces an intermediary process that detects holes in the depth image and uses corresponding color image data to fill or compensate for these missing regions. This intermediary step allows the system to maintain both 3D motion estimation capability and high recognition accuracy
2Measurement precision
If hole detection and compensation methods are applied, then object recognition accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent performs preliminary hole detection and compensation before the main object recognition process. By identifying and filling missing depth data in advance, the subsequent recognition algorithms can operate on complete data without requiring complex handling of missing values, thus improving accuracy without proportionally increasing overall complexity
Solution Approach 2:
The patent segments the processing into distinct stages: depth image processing, hole detection, color image integration, and feature extraction. This segmentation allows each stage to be optimized independently, managing complexity while achieving high recognition accuracy
Data Source
AI summary
An apparatus recognizes an object using a hole in a depth image. An apparatus may include a foreground extractor to extract a foreground from the depth image, a hole determiner to determine whether a hole is present in the depth image, based on the foreground and a color image, a feature vector generator to generate a feature vector, by generating a plurality of features corresponding to the object based on the foreground and the hole, and an object recognizer to recognize the object, based on the generated feature vector and at least one reference feature vector.