Abstract
In order to perform precise identification and location of artificial coded targets in natural scenes, a novel design of circle-based coded target and the corresponding coarse-fine extraction algorithm are presented. The designed target separates the target box and coding box totally and owns an advantage of rotation invariance. Based on the original target, templates are prepared by three geometric transformations and are used as the input of shape-based template matching. Finally, region growing and parity check methods are used to extract the coded targets as final results. No human involvement is required except for the preparation of templates and adjustment of thresholds in the beginning, which is conducive to the automation of close-range photogrammetry. The experimental results show that the proposed recognition method for the designed coded target is robust and accurate.
Introduction
Close-Range Photogrammetry (CRP) is a critical technique for establishing coordinate systems and measuring 3D geometries from 2D imagery. Central to this process is the use of artificial "coded targets." Most modern targets utilize circular geometries because their projection onto a camera sensor forms an ellipse, which possesses well-defined mathematical properties for localization.
However, traditional extraction algorithms rely heavily on parametric ellipse fitting. This approach represents a significant vulnerability: when targets are distant or viewed from oblique angles, the resulting image often lacks sufficient edge pixels for accurate parameter calculation. Furthermore, image noise and discretization errors can degrade fitting performance, leading to failures in the 3D reconstruction pipeline. This paper presents a robust, automatic system designed to overcome these limitations in complex natural scenes.
Methodology
The Target Design
The proposed planar target is bifurcated into two functional sections to decouple localization from identification:
- Target Box (Top): This section contains concentric circles—a white inner circle, a black outer circle, and a white ring—enclosed within a black rectangle. A critical feature is a small black indentation in the white ring, which breaks the rotational symmetry to provide orientation information.
- Coding Box (Bottom): This consists of a grid of eight blocks representing an 8-bit binary number. By separating the identification code from the localization feature, the system enhances overall robustness.
Extraction Algorithm
The extraction process follows a "coarse-to-fine" strategy that avoids the pitfalls of traditional ellipse fitting:
- Template Preparation: A library of templates is generated by applying geometric transformations—specifically scaling, rotation, and horizontal scaling—to the original target design. This accounts for various perspective distortions encountered in the field.
- Shape-Based Matching: The algorithm employs shape-based template matching to identify candidate regions within the image that match the pre-calculated templates.
- Center Point Extraction: Upon detection, Otsu’s adaptive segmentation and a region-growing algorithm are used to determine the exact center of the target box with sub-pixel precision.
- Parity Check: To eliminate false positives in complex backgrounds, the 8-bit code is read from the coding box. An 8th check bit (verified via bit-wise XOR) ensures data validity.
Experimental Results
The system was evaluated using high-resolution images captured in complex natural environments to test real-world applicability.
Accuracy and Speed
The algorithm demonstrated exceptional precision, with feature point localization deviations ranging from 0.158 to 0.281 pixels, even when subjected to heavy Gaussian noise. Efficiency is another hallmark of the approach; the entire processing pipeline for a high-resolution image takes approximately 0.2 seconds, making it viable for real-time engineering applications.
Robustness
The method proved resilient to perspective distortion, successfully identifying targets at viewpoint angles of up to 60°. The separation of the target and coding boxes, combined with the parity check, significantly reduced false alarm rates in cluttered industrial and outdoor scenes.
Conclusion
This research introduces a novel target design and a shape-based extraction algorithm that significantly improves the automation and reliability of close-range photogrammetry. By moving away from parametric ellipse fitting toward template matching and decoupled coding, the system achieves high accuracy and robustness in challenging environments. Future work will focus on further computational optimization and reducing the need for initial threshold adjustments to achieve full end-to-end automation.
Summary updated