In order to address the difficulty of accurate segmentation of froth images of different sizes, a method of froth image segmentation based on highlight correction and parameter adaptation is proposed. First, a machine vision system on a single-cell flotation machine is built to collect froth images. Homomorphic filtering is used to improve the uneven brightness and shadow of the images. Fuzzy c-means (FCM) clustering is then utilized to classify similar highlights that belong to the same froth. After Otsu threshold segmentation, a parameter-adaptive morphological operation is used to extract the marker points and edge bands and correct the froth edges in the original image. Finally, the modified image is filtered by morphological reconstruction, and the highlight mark is used as the local minimum point for watershed segmentation. Three sizes of froth images are segmented in comparative experiments. The results show that the proposed method is suitable for the segmentation of froth images of different sizes. The position of the extracted segmentation line is close to reality, with average over-segmentation and under-segmentation rates for froth images of 2.6% and 6.8%, respectively. The froth image segmentation performance is stronger than that of the other methods examined.

中文翻译：

---

基于高光校正和参数自适应的浮选泡沫图像分割

针对不同尺寸泡沫图像难以准确分割的问题，提出了一种基于高光校正和参数自适应的泡沫图像分割方法。首先，在单细胞浮选机上建立机器视觉系统来收集泡沫图像。同态滤波用于改善图像的不均匀亮度和阴影。然后利用模糊 c 均值 (FCM) 聚类对属于同一泡沫的相似亮点进行分类。Otsu阈值分割后，采用参数自适应形态学操作提取标记点和边缘带，校正原始图像中的泡沫边缘。最后对修改后的图像进行形态学重建滤波，将高亮标记作为局部极小点进行分水岭分割。在比较实验中分割了三种尺寸的泡沫图像。结果表明，该方法适用于不同尺寸泡沫图像的分割。提取的分割线位置接近真实，泡沫图像的平均过度分割和欠分割率分别为 2.6% 和 6.8%。泡沫图像分割性能强于其他检测方法。