Object Recognition Device Distance-Based Template Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition techniques using stereo cameras face challenges in accurately detecting objects at varying distances due to the lack of a suitable algorithm for both close and far objects, leading to potential inaccuracies in detection.
Innovation Solution
The implementation of an image processing device and object recognizing system that utilizes a parallax value computing unit and recognizing unit to enhance object detection accuracy by deriving parallax values and performing block matching processing, allowing for precise distance measurement and object recognition across different distances.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single template matching algorithm is used for all objects, then the device complexity is reduced, but the object detection accuracy deteriorates for objects at different distances
Solution Approach 1:
The patent applies parameter changes by switching between different template matching algorithms based on the distance to the object. For close objects, a first template matching algorithm with higher sensitivity to variations is used, while for far objects, a second template matching algorithm optimized for distant detection is applied. This dynamic adjustment of algorithm parameters resolves the contradiction by adapting the detection method to the specific distance condition, thereby maintaining high detection accuracy across varying distances without permanently increasing system complexity.
2Measurement precision
If different algorithms are used for close and far objects, then the object detection accuracy is improved, but the device complexity increases
Solution Approach 1:
The system implements self-service by automatically selecting the appropriate template matching algorithm based on the detected object distance. The distance information, obtained from the stereo camera system, serves as the decision criterion for algorithm selection. This self-adaptive mechanism eliminates the need for manual intervention or complex external control systems, thereby achieving high detection accuracy for both close and far objects while keeping the overall system complexity manageable through automated decision-making.
3Extent of automation
If template matching is used for tracking objects, then the tracking capability is achieved, but the detection accuracy deteriorates when object variation is large (close objects)
Solution Approach 1:
The patent applies dynamics by making the template matching process adaptive rather than static. The system dynamically switches between a first template matching algorithm for close objects with large variations and a second algorithm for far objects with small variations. This dynamic adaptation allows the tracking system to maintain accuracy despite changes in object appearance due to distance, thereby resolving the contradiction between automated tracking capability and detection accuracy for varying object conditions.
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 solution enhances object detection accuracy by effectively handling variations in object distance, enabling improved recognition of both close and far objects through the use of parallax value computation and block matching processing.
Implementation Method 1
imaging units 10a and 10b that take images of objects
Implementation Method 2
parallax value computing unit 3 that derives, from a plurality of images taken by the imaging units 10a and 10b, a parallax value dp representing a parallax with respect to an object
Data Source
Figure 1(a)~1(b)
Figure 2
Figure 3
AI summary
An information processing device includes a predicting unit configured to, from the position of an object in the previous frame with respect to the current frame, predict the position of the object in the current frame and identify a prediction area; a determining unit configured to determine whether the object is present in a first distance area or in a second distance area which is farther than the first distance area, based on the distance to the object in the previous frame; a first matching unit configured to perform template matching in the prediction area in the current frame using a first template regarding the object in the previous frame, to detect the object, when it is determined that the object is present in the first distance area; and a second matching unit configured to perform template matching in the prediction area in the current frame using a second template which is different than the first template and which is regarding the object in the previous frame, to detect the object, when it is determined that the object is present in the second distance area.