基于形状特征的XBT探头检测与识别研究

Study on XBT Probe Detection and Recognition Based on the Shape Feature

为改进国产投弃式温深仪(XBT)的下降速率公式,拟采用基于图像处理的方法测量其下落过程中的运动参数并分析,为此要实现XBT探头的自动提取。为实现视频图像中XBT探头的自动检测识别,提出了一种基于形状特征的目标检测识别算法。首先通过背景差分法去除背景噪声并阈值分割,采用连通区域的标记算子将二值图像标记为互不连通的待识别区域。然后提取连通区域的偏心率、致密性以及标记图构造描述目标的特征向量,通过计算各连通区域的特征评价函数,利用模糊综合评价方法识别XBT探头。最后对研究方法进行了物理实验验证。实验结果表明:用模糊综合评价方法能较好地在复杂背景下实现目标的识别,能够有效地在序列图像中准确的检测识别出XBT探头目标,为XBT运动参数的测量打下了基础。

In order to improve the fall rate equation of domestic-made XBT, it is necessary to measure and analyze its movement parameters based on image processing. To extract the targets of XBT probes accurately and rapidly from the complex background, a novel method for auto-detection and auto-recognition is proposed based on the shape feature. After background noises are eliminated by background subtraction, the appropriate threshold is selected and the binary images of difference images are obtained. The connected components of a binary image are labeled and a binary image is divided into several independent regions. On the basis of each region ＇s eccentricity, compactness and signature, an overall fuzzy evaluating method is adopted to recognize the XBT probes by calculating each region ＇s characteristic function. Then the method is validated through physical experiments. The results demonstrate that the target of XBT probes could be well detected and recognized by the overall fuzzy evaluating method under complex background, and this method can effectively extract XBT probes from image sequences, providing a theoretical and experimental foundation for the measurement of XBT movement parameters.