摘要
针对尿沉渣图像的背景和目标区分度低、有形成分复杂,导致提取困难的问题,设计了基于加权梯度和Snake模型的尿沉渣提取算法。首先采用基于区域生长的阈值分割方法对尿沉渣进行粗定位,然后通过形态学的方法确定Snake模型的初始蛇,最后将加权梯度融入到Snake模型中,完成对尿沉渣的准确提取。实验表明算法性能稳定,所提取的尿沉渣区域定位准确。特别是解决了虚假边缘和边缘断裂现象较为严重的尿沉渣轮廓提取。
In order to solve the problem that urine sediment can＇t be extracted effectively because of the poor discrimination between object and background and complex components in image, an algorithm of urine sediment extraction based on weighted gradient and Snake model was designed. Firstly, the approximate location of urine sediment was got by a region growing-based threshold segmentation method. Secondly, the preliminary snake of Snake model nearby the designed sediment contour was established with an algorithm based on morphology. Lastly, the weighted gradient was fused to Snake model so as to implement the accurate extraction of urine sediment. Experiment shows that this method has superior contour accuracy in localisation of urine sediment region extracted and stable performance, especially, it resolves the extraction of urine sediment contours with fake edges and break edges.
出处
《计算机应用与软件》
CSCD
2009年第4期100-102,108,共4页
Computer Applications and Software
基金
广西教育厅科研基金项目(200508107)
