D-S证据理论融合多特征的油菜病害识别方法

发布时间：2018-02-26 23:14

本文关键词： 多特征欧氏距离 D-S证据理论方差　出处：《电子测量与仪器学报》2017年01期 　论文类型：期刊论文

【摘要】：针对单一特征在识别油菜病害上存在的局限性,提出一种基于D-S证据理论融合多特征的油菜病害识别方法。首先对预处理后的油菜图片提取颜色矩、颜色共生矩阵两种特征,通过欧氏距离来构建D-S证据理论所必需的基本概率分配(BPA),最后运用D-S证据组合规则进行决策级融合,依据决策条件输出最终分类识别结果。针对存在最终识别结果被误识别为不确定问题,通过引入方差来对决策方法进行改进,避免了这一现象的产生。利用该方法在采集到的油菜样本上进行实验,取得了97.09%的识别率。实验表明,该方法能有效提高油菜病害识别率。
[Abstract]:Aiming at the limitation of single feature in rapeseed disease identification, a new method of rapeseed disease identification based on D-S evidence theory fusion was proposed. Firstly, the color moment and color-co-occurrence matrix were extracted from the pre-treated rape image. The basic probability distribution necessary for D-S evidence theory is constructed by Euclidean distance. Finally, the decision level fusion is carried out by using D-S evidence combination rule. According to the decision condition, the final classification recognition result is outputted. In view of the problem that the final recognition result is misidentified as uncertainty, the variance is introduced to improve the decision making method. The method was used to test the rape samples and the recognition rate of 97.09% was obtained. The experiment shows that the method can effectively improve the recognition rate of rape diseases.
【作者单位】：合肥工业大学计算机与信息学院;合肥工业大学情感计算与先进智能机器安徽省重点实验室;
【基金】：安徽省自然科学基金(1408085MKL16) 国家自然科学基金(614320040,61672202) 国家自然科学青年基金(61300119,61502141)资助项目
【分类号】：S435.654;TP391.41

【相似文献】