Distance Detection Device Using Asynchronous Stereo and Viterbi Algorithm
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional distance detection devices using pattern matching for stereo cameras have a high processing load, making them inefficient for accurately detecting distances to objects like preceding vehicles from captured images.
Innovation Solution
A distance detection device employing a pixel cost calculation, minimum cost pixel calculation, corresponding point setting, object extraction, and distance calculation sections, utilizing the Viterbi algorithm and asynchronous stereo method to calculate parallax costs and determine object coordinates with reduced processing load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pattern matching is used for image processing in stereo cameras, then object detection accuracy is improved, but processing load increases significantly
Solution Approach 1:
The patent segments the image processing task into distinct stages: feature point extraction from reference image, feature point extraction from comparative image, and matching between corresponding points. This segmentation allows each stage to use optimized algorithms appropriate to its specific function, reducing overall processing load while maintaining detection accuracy.
Solution Approach 2:
The patent extracts only salient feature points from images rather than processing all pixels. By identifying and extracting key feature points (corners, edges, distinctive patterns) from both reference and comparative images, the system reduces the data volume requiring complex matching operations, thereby lowering processing load while preserving object detection accuracy.
2Measurement precision
If complex image processing algorithms are applied, then detection accuracy is improved, but device complexity increases
Solution Approach 1:
The processing structure is segmented into modular sections: a reference image processing section, a comparative image processing section, and a matching section. Each module performs a specific function with a dedicated algorithm, making the overall system easier to implement and maintain while achieving high detection accuracy through coordinated operation of these simplified modules.
Solution Approach 2:
The patent introduces feature points as intermediary elements between the raw images and the final distance calculation. These feature points serve as mediators that capture essential image information in a compact form, enabling accurate matching without requiring complex direct image-to-image comparison algorithms.
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
Enables accurate detection and recognition of objects with a simple processing structure, reducing the computational burden and improving detection accuracy for distance calculation.
Implementation Method 1
a parallax cost representing a cost regarding a change amount of parallax. The parallax represents a coordinate difference between a reference pixel in a reference image and a comparative pixel in a comparative image
Data Source
AI summary
A distance detection device calculates a pixel cost, every reference pixel, based on a difference between reference pixel information in a reference image and comparative pixel information in a comparative image while switching the reference and comparative images. The device calculates a parallax cost, every reference pixel, representing a cost regarding a change-amount of the parallax as a coordinate difference between a reference pixel and a comparative pixel when the reference image is switched. The device calculates a combination of each reference pixel and a comparative pixel having a minimum total cost every reference pixel. The minimum total cost represents a sum of the pixel cost and the parallax cost. The device obtains a relationship of a corresponding point between each reference pixel and its corresponding comparative pixel, and calculates a distance to an object in each captured image based on the relationship of the corresponding point.