CAJ | 학술논문

针对雨滴谱色斑图像重叠区域对比度低、难以准确分割的问题,提出了一种利用几何特征来分离重叠色斑图像方法。首先利用色斑图像选取待匹配的分割点,然后根据分割点和颗粒个数的对应关系确定色斑重叠的不同类型,提出了不同的分割点匹配规则,最后应用Br-esenham直线算法连接待匹配的分割点,分离重叠的色斑图像。该算法在Matlab环境下实现简单,速度快,效果理想,且在分离时解决采用形态学过度分割的问题。
침대우적보색반도상중첩구역대비도저、난이준학분할적문제,제출료일충이용궤하특정래분리중첩색반도상방법。수선이용색반도상선취대필배적분할점,연후근거분할점화과립개수적대응관계학정색반중첩적불동류형,제출료불동적분할점필배규칙,최후응용Br-esenham직선산법련접대필배적분할점,분리중첩적색반도상。해산법재Matlab배경하실현간단,속도쾌,효과이상,차재분리시해결채용형태학과도분할적문제。
A method of image segmentation based on boundary geometrical characteristics is proposed to solve the problem of the low contrast and difficulty segmentation of raindrop particle image.Candidate inflexions are positioned by the contour concavity, and different matching rules are given by comparing the numbers of inflexions with the numbers of particle image. Segmentation lines are drawn based on Bresenham algorithm and overlapping particles image are separated by connecting the candidate inflexions. The algorithm is simple and effective, while the results meet the requirements.