【作者】 陆宗骐； 陈芳； 王学城；

【Author】 Lu Zong-qi, Chen Fang, Wang Xue-cheng(College of Information, East China University of Science and Technology, Shanghai 200237;

【机构】 华东理工大学信息学院；

【摘要】 本文在原来的Laplacian算子计算结果中增加偏移量offset并仅取结果中的正值(与零)部分,以达到检测图象中的点与线条的目的。当对象比背景亮时,offset取1,用于检测明区边缘,为黑底白线;当对象比背景暗时,iffset取254,用于检测暗区边缘,为白底黑线。后者类似画家所作的素描,可用于人物素描的制作。图象处理中,此算法可直接用于检测粒子和单点宽的线条,与线条掩模相结合也可检测较粗的线条。文中给出了在DPIV(数字粒子测速技术)、焊缝和菌丝图象中检测粒子与线条的实例。本文还提出了根据单峰直方图下降段的波形分析确定分割阈值的方法。更多还原

【Abstract】 In this paper, an offset is added to the result of Laplacian operator and only positive values and zero can be chosen. In this way, both dots and lines can be detected. The offset is selected as 1 to detect the edge of bright area when object is brighter than background, otherwise, offset is selected as 254 to detect the edge of dark area when object is darker than background. The latter is similar to pencil sketch made by painter which can be used to make profile. In image processing, the operator can be used to detect particles and lines directly; it can also be used with line mask to detect thick lines. Examples of particle and line detection in DPIV(Digital Particle Image Velocimetry), welding line and hypha image are proposed in this paper. And also a method to define the threshold is found according to the down part of wave of single peak histogram.更多还原