【摘要】 为了识别烟叶图像中的异物,利用文献[1]中提出的颜色模型建立方法,为烟叶建立烟叶颜色模型。针对该模型可以得到具有二维属性特征的样本数据,使用支持向量数据域描述的方法(SVDD)训练样本数据,得到特征数据的一个包围域从而得到分类器模型。为了提高单幅异物图像的识别速度,对分类模型建立查找表,从而识别出异物,经验证本方法识别效果颇佳,并且对于和烟叶颜色相近的异物的识别也有所提高,综合剔除率高达98.3%,足以满足实际的剔除设备的要求。更多还原

【Abstract】 In order to recognize the foreign body in the Tobacco image,this paper establishes tobacco color model use the distribution of tobacco color space base on hierarchical clustering,proposed by Gong Wei,Wu Tiejun.The feature data with two-dimensional can be obtained by using this tobacco model,and use the method of support vector data field(SVDD) to train the feature data,finally can obtain the classifier model.In order to improve the speed of recognizing foreign object image,this paper establishes a look-up table to identify the foreign body,the effect of this method to identify tobacco foreign body is quite good,and the foreign body color is similar to tobacco color has also improved.The recognition rate is up to 98%,enough to meet the actual demand of equipment.更多还原