基于模糊自适应Hough变换的海洋环境特征提取

Sea environment feature extraction based on fuzzy adaptive Hough transform

针对海洋环境中特征提取难度大、效率低等问题,设计了基于模糊自适应Hough变换的海洋环境特征提取方法。通过建立成像声呐观测模型,对声呐数据进行预处理并剔除野值;将声呐坐标系下的测量数据转化为全局坐标系下,并提取环境的全局特征;根据梯度方向信息,模糊化处理声呐数据点,采用极小极大模糊推理评判数据点属于某条直线特征的可能性,自适应选择参与投票的数据点。基于海试数据的试验结果表明:该方法能够从伴有噪声、混响、反射的声呐数据中准确地提取港口环境的线特征,并且模糊自适应Hough变换特征提取的投票数目与计算时间分别为传统Hough变换的32.5%和26.1%,表明模糊自适应Hough变换具有存储空间小、计算效率高、实用性强等优点。

Aiming at the high difficulty and low efficiency in sea environment feature extraction, a sea environment feature extraction method based on fuzzy adaptive Hough transform is designed. By establishing the measurement model of imaging sonar, sonar data is preprocessed to eliminate the outliers. The measurement data under sonar coordinate system is transformed into global coordinate system to extract global features. The sonar data points are processed with fuzzy reasoning based on gradient orientation information. The maximum and minimum fuzzy reasoning principle is adopted to evaluate the probability that the data point belongs to a line. The data points that participate in voting are selected adaptively. Experiment result based on sea trial data shows that the method could accurately ex- tract line features of harbor environment from the sonar data with noise, reverberation and reflection. The number of votes and computational time of feature extraction using fuzzy adaptive Hough transform are 32.5 % and 26.1% of those using traditional Hough transform, respectively ,which shows that the fuzzy adaptive Hough transform possesses the advantages of small storage space, high computational efficiency and strong practicality.