Chan-Vese模型两相自适应窄带的阴影区检测方法  

Shadow regions detection algorithm by adaptive narrowband two-phase Chan-Vese model

作　　者：王兴梅[1] 印桂生[1] 刘光宇[2] 刘志鹏[1] 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,哈尔滨150001 [2]哈尔滨工程大学船舶工程学院,哈尔滨150001

出　　处：《声学学报》2015年第6期816-826,共11页Acta Acustica

基　　金：国家自然科学基金(41306086);哈尔滨市基金(2014RFQXJ105);中央高校基本科研业务费专项(HEUCFR1121;HEUCF100606)资助

摘　　要：为得到快速高精度的声呐图像阴影区检测效果,提出Chan-Vese模型两相自适应窄带检测方法。利用各向异性二阶邻域马尔可夫模型估计声呐图像的纹理特征参数,实现原始图像平滑去噪;由块方式的k-均值聚类算法确定图像的初始两类分割,初步确定阴影区大致位置,并根据此大致位置,自适应初始化零水平集函数,来减少人为干预,提高检测速度;在此基础上,提出建立Chan-Vese模型两相窄带水平集进行声呐图像检测,完成局部寻优,排除全局图像中孤立区对检测的影响,使阴影区检测结果更加精确。通过对真实声呐图像的检测实验结果分析,验证提出的检测方法能够去除原始图像的部分噪声,提高检测精度和速度,有一定的自动性和适应性。To obtain fast and highly accurate shadow regions detection results of sonar image, a detection algorithm of adaptive narrowband two-phase Chan-Vese model is proposed in the paper. The anisotropic second-order neighborhood MRF （Markov Random Field, MRF） is used to describe texture feature parameters of sonar image, and complete the noise smoothing. Initial two-class segmentation is determined by the block mode k-means clustering algorithm, to preliminarily estimate the approximate position of the shadow regions. Then, in order to reduce human intervention and improve the detection speed, zero level set function is adaptively initialized by approximate position of shadow regions. On this basis, the narrowband level set of two-phase Chan-Vese model is proposed to detect the sonar image and complete local optimization, which eliminates the global image＇s interference in detection results, makes shadow regions detection results more accurate. Experimental results demonstrate that the proposed algorithm can remove partial noise, improve the detection speed and accuracy, it has certain automaticity and adaptability.

关 键 词：CHAN-VESE模型 图像检测 阴影区 自适应 窄带 K-均值聚类算法 声呐图像 原始图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]