基于凹凸变换与圆周拟合的重叠气泡轮廓重构

发布时间：2018-05-01 02:37

本文选题：凹凸性变换 + 圆周拟合　；参考：《浙江大学学报(工学版)》2017年04期

【摘要】：针对识别气液两相流的气泡过程中气相比例较大气泡易发生重叠的问题,提出基于曲线凹凸性变换与圆周拟合的重叠气泡分割与轮廓重构算法.对气泡图像进行预处理,利用高斯滤波滤除噪声干扰,依据自适应阈值分离出前景,得到预分割图像.利用Canny算法提取轮廓,利用高斯函数平滑曲线,计算轮廓上的点到质心的距离.对得到的距离曲线进行凹凸性变换特性分析,对每两个凹点进行分割路径选取.利用最小二乘圆周拟合插值算法对重叠部分的气泡轮廓进行重构,得到气泡分割后的重构图像.实验结果表明,该算法是一种有效的分割与重构重叠气泡的算法,具有较好的分割和重构性能.
[Abstract]:In order to solve the problem that large gas bubble ratio is easy to overlap in the process of identifying gas-liquid two-phase flow, an overlapping bubble segmentation and contour reconstruction algorithm based on curve convexity transformation and circumference fitting is proposed. The bubble image is preprocessed, and the noise interference is filtered by Gao Si filter. The foreground is separated according to the adaptive threshold, and the presegmented image is obtained. Canny algorithm is used to extract the contour and Gao Si function is used to smooth the curve to calculate the distance between the points on the contour and the center of mass. The characteristic of convexity transformation of the distance curve is analyzed, and the path of each two concave points is selected. The bubble contour of the overlapped part is reconstructed by using the least square circumferential fitting interpolation algorithm, and the reconstructed image is obtained after the bubble segmentation. The experimental results show that this algorithm is an effective algorithm for the segmentation and reconstruction of overlapping bubbles, and has good performance of segmentation and reconstruction.
【作者单位】：浙江大学流体动力与机电系统国家重点实验室;
【基金】：国家自然科学基金资助项目(51405435,51405434) 浙江省公益性技术应用研究计划资助项目(2016C31035) 浙江省自然科学基金资助项目(LY17F030011)
【分类号】：TP391.41;O359

【相似文献】