Fiducial Marker Detection Using Dual-Level Binary Codes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fiducial marker systems are susceptible to lighting conditions and require sophisticated computer vision for detection, often failing due to binarization methods and insufficient information encoding, which limits their robustness and practical usability in real-world applications.
Innovation Solution
A fiducial marker system using a polygonal border with dual-level binary digital code, incorporating checksums and error correction, detected through an edge-based method that accounts for perspective distortion and occlusion, allowing for reliable detection across varying lighting conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If binary thresholding algorithms are used to classify image pixels for marker detection, then the marker system can be simpler and less expensive in hardware requirements, but the system becomes susceptible to lighting conditions and fails under varying illumination
Solution Approach 1:
The patent changes the detection parameter from intensity-based binary thresholding to gradient-based edge detection. By using Sobel operators to compute gradient magnitudes and directions, the system becomes insensitive to absolute intensity levels and adapts to varying lighting conditions while maintaining simple hardware requirements.
Solution Approach 2:
The patent applies preliminary gradient computation and edge detection before marker identification. By pre-processing the image to enhance edges and compute gradients, the system prepares the data in a form that is robust to lighting variations, enabling reliable detection without complex hardware.
2Reliability
If sophisticated computer vision systems are used to detect passive markers, then detection reliability can be improved, but the system complexity and computational requirements increase
Solution Approach 1:
The patent segments the marker detection process into distinct stages: gradient computation, non-maximum suppression, edge tracing, and marker identification. This segmentation allows each stage to be optimized independently, achieving high reliability through systematic processing while keeping individual computational steps simple and efficient.
Solution Approach 2:
The patent replaces complex intensity-based classification mechanisms with gradient-based edge detection mechanics. By substituting the detection approach from direct pixel intensity analysis to gradient computation and edge tracing, the system achieves robust reliability with simpler, more efficient computational operations.
3Ease of manufacture
If binary patterns are used in markers, then the markers can be printed with standard printers and are easier to manufacture, but the patterns are not always effective under varying lighting conditions
Solution Approach 1:
The patent changes the detection parameter from absolute intensity values to gradient magnitudes and directions. This parameter transformation allows standard binary printed patterns to be effectively detected under varying lighting conditions, as gradients capture local intensity changes rather than absolute levels.
Solution Approach 2:
The patent effectively uses 'gradient changes' analogous to color changes by computing gradient magnitudes and directions. This transforms the detection approach to rely on directional intensity changes in the pattern, making binary printed markers robust to illumination variations while maintaining ease of manufacture.
4Adaptability or versatility
If existing fiducial marker systems are used, then the system can be implemented with available technology, but false positive and false negative rates are high due to susceptibility to lighting and occlusion
Solution Approach 1:
The patent incorporates feedback through non-maximum suppression and edge tracing validation. The system suppresses non-maximal gradient responses and validates edge continuity, providing feedback mechanisms that reduce false positives and false negatives while working with available camera technology.
Solution Approach 2:
The patent applies preliminary gradient computation and edge detection before final marker identification. This preliminary processing filters out spurious detections and prepares robust features for identification, significantly reducing false positive and false negative rates while maintaining compatibility with standard technology.
Data Source
AI summary
This invention discloses marker detectable by visual means comprising; a polygonal border having of at least four non collinear salient points. The marker has on it a pattern in binary digital code. The binary code data contains information data, checksum and error correction. The binary code data is on two levels; a first level of binary code readable at a first given distance, a second level of binary code readable at a second given distance, where the second given distance is less than the first given distance and the second level binary code is smaller in size than the first level binary code. The second level of binary code does not interfere with the reading of the first level binary code. It also discloses a method for detecting a marker comprising the steps of detecting an image, using an edge detector to detect an edge in said image, grouping more than one edge into a polygon having salient points, calculating homography from polygon salient points, generating a list of homographies, extracting binary data from input image having homographies, identifying and verifying binary data.