基于计算机视觉的番茄营养元素亏缺的识别

Diagnosis of Nutrient Deficiency of Tomato Based on Computer Vision

以肉眼不易识别的番茄缺氮和缺钾初期为研究对象 ,对体现在叶片颜色和纹理上的缺素症状进行了特征提取 ,利用遗传算法对提取的众多缺素特征进行优化组合 ,选择出用于模式识别分类器设计的特征向量。建立了二叉树分类法对番茄缺素进行模式识别的框架 ,在该框架下 ,基于模糊 K近邻法建立了缺素的模式识别系统 ,并进行了识别测试。结果表明 ,对不易肉眼判别的番茄缺氮和缺钾初期叶片的识别准确率在 85 %以上 。

The characteristic features of nitrogen and kalium deficiencies for tomatoes, which is hardly to be recognized in the initial period of growing, mainly represents on the color and texture of tomato leaves. The extraction of these nutrient deficiency features was made and the extracted features were optimized and combined to pick the eigenvectors out for the design of a mode identifying assorter. A mode identifying frame for deficient elements was set up by the tree taxonomy with two branches and then a mode identifying system was established based on the K neighbour approach. The results of the mode identify showed that the accuracy of the diagnosis reached about 85% before the symptoms could be recognized by eye.