中文摘要：

裂隙特征作为膨胀土裂隙研究的基础,其定量化描述有助于进一步研究裂隙对膨胀土工程性质的影响。以膨胀土平面裂隙为研究对象,使用数码摄影获取膨胀土裂隙图像,通过优化和改进裂隙图像处理方法及裂隙特征提取方法,获得膨胀土裂隙的各特征参数。对于裂隙图像处理方法,采用改进的局部阈值二值化方法,相比大津法能更好地将裂隙和背景区分开来,而通过标记连通区域的方法有效去除了裂隙二值图像中全局分布的细小杂点,同时,利用闭运算填充了裂隙图像中细小的孔洞和缺口,上述处理方法使裂隙二值图像更清晰可靠。对于裂隙特征提取方法,基于对裂隙端点和交点的快捷识别能力,选择了查找表方法作为裂隙条数统计的基础;改进了端点识别算法以去除毛刺,提高了裂隙长度统计的精度;采用改进的最小外接矩形算法进行裂隙宽度的统计,提高了计算精度,节省了计算时间;尝试采用玫瑰花图进行裂隙方向统计,更直观明了地分析裂隙方向及条数的变化规律。

英文摘要：

As a basis of crack research of expansive soil, crack features of quantitative description contributes to further investigate the influence of cracks on engineering properties of expansive soils. Taking surface cracks of expansive soil as research objects, crack images are captured by using digital imaging, and parametes of crack features are obtained through optimizing and improving methods of crack image processing and crack feature extraction. For methods of crack image processing, it showed that improved local threshold binarization method could distinguish crack and background better than Ostu method. By marking connected areas, globally distributed noises on the crack binary image is eliminated effectively; and employing closing operation, tiny holes and gaps could be filled. The above methods make the crack binary image clearer and more reliable. For crack feature extraction methods, lookup table method is adopted to quickly identify crack endpoints and intersections, which helps count number of crack branches. Improved endpoints recognition algorithm was used to remove burrs from crack trunk, which makes the statistics of crack length more accurate. And improved minimum enclosing rectangle algorithm is used for statistics of crack width, which improves the calculation accuracy and saves the calculation time. Rose diagram is employed for statistics of crack orientation, which could be more intuitive to analyze the change law of crack orientation and number.