Stereo Camera Object Recognition via Common Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems face limitations in resilience to environmental variations and require manual operation for feature extraction and registration, making them unsuitable for robust recognition in dynamic environments, especially for robots used in home services.
Innovation Solution
An object recognition system utilizing a stereo camera to extract texture-based features from left and right images, comparing feature vectors using Euclidean distance, and recognizing objects based on common features stored in a database, improving invariance performance and stability through one registration operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single camera is used for object recognition, then the system is simpler, but the recognition performance and invariance to environmental variation are limited
Solution Approach 1:
The patent merges left and right image data from a stereo camera system into a unified feature extraction process. By combining information from both cameras, the system achieves improved recognition performance and invariance to environmental variations while managing the complexity through integrated processing rather than separate independent systems.
Solution Approach 2:
The patent transitions from single-camera 2D image recognition to stereo-camera 3D spatial recognition. By utilizing the depth dimension provided by the stereo camera setup, the system extracts features that are invariant to environmental variations such as illumination changes and camera angle variations, thereby improving reliability.
2Adaptability or versatility
If manual operation is used for feature extraction and registration, then the process is controllable, but it is not suitable for robust recognition in dynamic environments and requires significant user intervention
Solution Approach 1:
The patent implements automatic feature extraction and registration by utilizing common features detected in both left and right images. The system self-adjusts to environmental variations by naturally selecting features that appear consistently across both stereo images, eliminating the need for manual intervention while maintaining robustness.
Solution Approach 2:
The system uses feedback from the stereo matching process to automatically adapt to environmental variations. By comparing features between left and right images and selecting common features, the system continuously adjusts its recognition process to maintain performance across varying conditions without user input.
3Adaptability or versatility
If a large number of images are collected for training, then the recognition can cope with object variation, but the registration process becomes troublesome and time-consuming
Solution Approach 1:
The patent extracts only the essential common features that are present in both left and right images, rather than requiring comprehensive training datasets. This selective extraction of discriminative features enables the system to handle object variation efficiently without the time-consuming process of collecting and registering large numbers of training images.
Data Source
AI summary
Disclosed herein are an object recognition system and method which extract left and right feature vectors from a stereo image of an object, find feature vectors which are present in both the extracted left and right feature vectors, compare information about the left and right feature vectors and the feature vectors present in both the extracted left and right feature vectors with information stored in a database, extract information of the object, and recognize the object.