改进的分水岭算法在牛乳体细胞上的分割及其应用

Segmentation and application of improved watershed algorithm on milk somatic cells

研究牛乳体细胞出现重叠和粘连问题时有效可行的分割方法。对原图像二值化处理,后对灰度梯度图像进行形态学平滑滤波处理,然后提取图像的前景标记和背景标记同时对梯度图进行修正,最后采取分水岭进行分割。仿真结果显示,分水岭算法可有效的抑制图像过分割的情况,且能准确、快速的分割牛乳体细胞,同时可为后续步骤牛乳体细胞的计数和诊断提供可靠的依据。

the effective and feasible segmentation method for overlapping and adhesion of bovine milk somatic cells was studied.The original image is binarized,then the gray-gradient image is processed by morphological smoothing filtering,then the foreground mark and background mark of the image are extracted and the gradient image is corrected.Finally,the watershed is used to segment the image.The simulation results show that the watershed algorithm can effectively suppress the image over-segmentation and can segment the bovine milk somatic cells accurately and quickly.At the same time,it can provide a reliable basis for the counting and diagnosis of bovine milk somatic cells in the subsequent steps.