基于分段面积比的植物叶片图像识别方法研究
|
摘要
为了实现对植物叶片图像的快速识别,以植物叶片分段面积比作为叶片的关键特征,构建植物叶片形态特征模型。先将植物叶片图像进行二值化、腐蚀、膨胀等常规化预处理,找到叶片图像最小外接矩形;然后对处理后的叶片图像进行面积等分分段处理,计算各段叶片面积与该段矩形面积的比值,形成叶片特征数据样本集。在此基础上,分别使用三层神经网络算法和多元线性回归算法,对18种植物叶片的特征数据集进行了识别实验。结果表明,运用这种方法进行植物叶片图像识别,运算简单,识别率高,易于实现;神经网络算法的识别效果明显优于多元线性回归算法。
In order to realize the fast recognition of plant leaves, the plant leaf morphological feature model is constructed by using the plant blade leaf area ratio as the key feature of the leaf. Firstly, the plant leaf image is subjected to conventional pretreatment such as binarization, corrosion and expansion, and then the minimum circumscribed rectangle of the leaf image is found. Then, the processed blade image is divided into several segments of the same size, and the ratio of the blade area of each segment to the rectangular area of the segment is calculated to form a sample set of the blade feature data. On this basis, the NN algorithm and the MLR algorithm are used to perform leaf recognition experiments on this new blade feature data set. This research method is applied to the identification of 18 plant leaves. The results show that the plant leaf recognition method based on the sectional area ratio is simple and good; the recognition effect of the NN algorithm is better than that of the MLR algorithm.
In order to realize the fast recognition of plant leaves, the plant leaf morphological feature model is constructed by using the plant blade leaf area ratio as the key feature of the leaf. Firstly, the plant leaf image is subjected to conventional pretreatment such as binarization, corrosion and expansion, and then the minimum circumscribed rectangle of the leaf image is found. Then, the processed blade image is divided into several segments of the same size, and the ratio of the blade area of each segment to the rectangular area of the segment is calculated to form a sample set of the blade feature data. On this basis, the NN algorithm and the MLR algorithm are used to perform leaf recognition experiments on this new blade feature data set. This research method is applied to the identification of 18 plant leaves. The results show that the plant leaf recognition method based on the sectional area ratio is simple and good; the recognition effect of the NN algorithm is better than that of the MLR algorithm.
引文
[1] 恩德,忽胜强.基于集成神经网络的植物叶片识别方法[J].浙江农业学报,2015,27(12):2225-2233.
[2] 刘骥,曹凤莲,甘林昊.基于叶片形状特征的植物识别方法[J].计算机应用,2016,36(S2):200-202.
[3] 高良,闫民,赵方.基于多特征融合的植物叶片识别研究[J].浙江农业学报,2017,29(4):668- 675.
[4] 马娜,李艳文,徐苗.基于改进SVM算法的植物叶片分类研究[J].山西农业大学学报(自然科学版),2018,38(11):33-38.
[5] 王礼,洪祖兵,方陆明,等.基于iOS系统的观赏植物识别[J].浙江农林大学学报,2018,35(5):900-907.
[6] 谢超凡,徐鲁雄,徐琳.基于神经网络的教学评分系统模型[J].内江师范学院学报,2016,31(2):8-11.
[7] 陈晖,胡泽根,王永平,等.广义回归神经网络技术在新油田快速评价中的应用[J].重庆科技学院学报(自然科学版),2019,21(1):42-45.
[2] 刘骥,曹凤莲,甘林昊.基于叶片形状特征的植物识别方法[J].计算机应用,2016,36(S2):200-202.
[3] 高良,闫民,赵方.基于多特征融合的植物叶片识别研究[J].浙江农业学报,2017,29(4):668- 675.
[4] 马娜,李艳文,徐苗.基于改进SVM算法的植物叶片分类研究[J].山西农业大学学报(自然科学版),2018,38(11):33-38.
[5] 王礼,洪祖兵,方陆明,等.基于iOS系统的观赏植物识别[J].浙江农林大学学报,2018,35(5):900-907.
[6] 谢超凡,徐鲁雄,徐琳.基于神经网络的教学评分系统模型[J].内江师范学院学报,2016,31(2):8-11.
[7] 陈晖,胡泽根,王永平,等.广义回归神经网络技术在新油田快速评价中的应用[J].重庆科技学院学报(自然科学版),2019,21(1):42-45.