一种改进的葡萄叶片自动分割算法

针对不同光照以及不同天气条件造成葡萄叶片图像不能准确分割的问题,提出了一种改进的图割分割算法。采用G/R以及a*颜色特征自动选择叶片目标和背景种子点,利用混合高斯模型对叶片和背景的概率密度分布进行估计;在马尔科夫随机场的基础上,建立像素特征的能量函数;通过求解能量函数最小化对叶片实现了自动分割。对多种不同分割特征的分割效果进行对比试验,结果表明:对于不同时间、不同天气的叶片图像,单一G/R和a*具有较好的效果,分割精度分别达到86.74%和92.38%,若用它们组合为双特征,分割效果会进一步提高,分割精度可达95.03%。

An improved automatic segmentation algorithm for grape leaves

In order to tackle the problem that grape leaf image can not be segmented accurately under different light and weather conditions.An improved graph segmentation algorithm is proposed.Firstly,the seed points are automatically selected by G/R and a^* color characteristics,and the probability density of the target and the background is estimated by using the mixed Gaussian model.Then based on the Markov random field,the energy function of pixel feature is established.Finally,the cuts are realized by solving the energy function minimization.The effects on segmentation by many different segmentation features are compared.The results shows that single use of G/R or a^* gives better result for leaves image in different time and weather.The segementation accuracy can reach 86.74% and 92.38%,respectively.If they are combined into double characteristic,the result of the segementation will be more accurate and the segmentation accuracy can reach 95.03%.