Shi Baoquan, He Qian, Chen Xianmin, Zhang Wendong, Yang Lin
School of Mechano-Electronic Engineering, Xidian University, Xi'an 710071, China.
National Key Laboratory of Strength and Structural Integrity, Xi'an 710065, China.
Sensors (Basel). 2025 Mar 28;25(7):2158. doi: 10.3390/s25072158.
Circular dots are widely used in various measurement applications due to their inherent symmetry, ease of detection, and scalability. However, when degraded by factors such as specular highlights, low contrast, strong noise, or friction damage, accurately extracting them from the background becomes a significant challenge. To address this issue, an enumeration-based thresholding method (EBTM) is proposed for degraded circular dot segmentation. Firstly, a series of candidate outputs are generated using an enumeration-based thresholding scheme. Next, an assessment criterion is developed to evaluate these candidate outputs. Finally, the optimal segments are selected from each candidate output and combined to produce a reasonable thresholding result. Unlike traditional methods, the novel approach does not focus on selecting the optimal threshold values, but instead aims to choose the best segments to produce the desired output. Owing to the enumeration-based thresholding mechanism, the novel approach demonstrates greater robustness in handling the challenges in degraded circular dot images. Extensive comparative studies demonstrate the superiority of the novel approach.
圆形点由于其固有的对称性、易于检测和可扩展性，在各种测量应用中被广泛使用。然而，当受到镜面高光、低对比度、强噪声或摩擦损伤等因素影响而退化时，从背景中准确提取它们就成为一项重大挑战。为了解决这个问题，提出了一种基于枚举的阈值化方法（EBTM）用于退化圆形点分割。首先，使用基于枚举的阈值化方案生成一系列候选输出。接下来，制定一个评估标准来评估这些候选输出。最后，从每个候选输出中选择最优片段并进行组合，以产生合理的阈值化结果。与传统方法不同，这种新方法不侧重于选择最优阈值，而是旨在选择最佳片段以产生所需输出。由于基于枚举的阈值化机制，这种新方法在处理退化圆形点图像中的挑战时表现出更强的鲁棒性。广泛的比较研究证明了这种新方法的优越性。
