| 摘要: |
| 生物发光现象在海洋中广泛存在，基于微光成像技术对深海发光浮游生物的发光现象进行视觉捕获是一种先进的原位观测手段，但目前缺乏有效的方法对获取的生物发光影像资料进行自动分析。针对此问题，本文提出一种基于多特征匹配跟踪的发光浮游生物自动计数方法，基于同一目标在连续帧间的运动具有连续性的思想，首先采用帧间质心差小于一定阈值的原则进行初始粗匹配，然后针对目标过近造成的误匹配问题再进一步进行运动方向的匹配，提高帧间匹配精度，从而准确计数。采用所提出的方法对深海发光浮游生物的真实视频数据进行了自动分析，实验结果证明该计数方法具有较好的准确率，对生物发光区域的尺度变化、发光目标间距过近等情况具有较好的鲁棒性，能达到较好的计数效果。 |
| 关键词: 深海发光浮游生物 多特征匹配 跟踪计数 |
| DOI：10.11759/hykx20181226004 |
| 分类号:TP-391 |
| 基金项目:国家重点研发计划重点专项课题（2016YFC0302402） |
|
| Automatic counting method for deep-sea luminescent plankton based on multifeature match tracking |
|
XU Jiao1, ZHAO Qi-jie1,2, ZHANG Jun-shao1, ZHANG Xi1
|
|
1.School of Mechatronics Engineering and Automation, Shanghai University, Shanghai 200072, China;2.Shanghai Key Laboratory of Intelligent, Shanghai 200072, China
|
| Abstract: |
| The bioluminescence phenomenon is widespread in the ocean, and the visual capture of the bioluminescence phenomenon in deep-sea luminescent plankton through low-light imaging technology is an advanced in situ observation method. However, there is no effective method to analyze the acquired bioluminescence imaging data automatically. To address this problem, an automatic counting method for deep-sea luminescent plankton based on multifeature match tracking was proposed in this paper. This method is mainly based on the idea that the same target exhibits continuous motion between successive frames. First, the initial coarse matching is performed on the basis of the principle that the inter-frame centroid difference is less than a certain threshold. Furthermore, the matching of the motion direction is performed to address the mismatch problem caused by targets that are too close. Thus, the inter-frame matching precision is improved, thereby achieving accurate counting. The proposed method is used to analyze the real-time video data of deep-sea luminescent plankton automatically. The results show that the automatic counting method exhibits high accuracy and is robust to the scale change of the bioluminescent region and the close distance of the illuminating target, thereby achieving good counting results. |
| Key words: Deep-sea luminescent plankton multi-feature tracking to count |
