摘要
针对路面图像对比度低、背景纹理复杂、噪声多的特点,首先通过设计一种频域滤波器进行滤波,去除部分噪声,增强裂缝图像和抑制背景的干扰;其次基于多尺度分析的优点,选取不同尺寸的结构元素进行形态学运算,进一步去除噪声及伪裂缝;最后进行Beamlet变换,对不同尺寸、不同位置和方向的裂缝目标进行精确逼近,重构裂缝图像从而提取目标。试验结果证明了该方法的有效性,取得了比较好的效果。
In view of the characteristics of the low contrast of the image on the road surface, complicated texture of background and much noise, firstly, a frequency domain filter is designed for filtering to remove part of the noise and enhance crack image and restrain the interference of background; secondly based on the advantages of multi-scale analysis, different sizes of structural elements are chosen for morphological calculations to further remove noise and pseudo crack; finally Beamlet transform is conducted to try precisely as much as possible extract the cracks with different sizes, different positions and different directions, and re-construct crack image to extract target detection. The experimental results prove that the proposed method is effective and has achieved good effect.
出处
《公路》
北大核心
2018年第1期31-34,共4页
Highway
基金
江西省教育厅科技资助项目,项目编号GJJ13373
国家自然科学基金,项目编号61463020
