摘要
利用感知哈希算法,针对水稻不同生长发育期的特征,通过比较图像相似度实现水稻生长发育期的自动识别。通过感知哈希算法提取每张待测图片的哈希码,并与已标注过发育期的图库中的水稻图像哈希码进行对比,找出相似度最高的哈希码所代表的标签,输出水稻的移栽期、返青期、分蘖期、拔节期、孕穗期、抽穗期、乳熟期、成熟期共8个主要生长发育期,并针对该算法进行了鲁棒性、敏感性和检索准确性试验。将自动识别的结果与气象观测站台的人工观测数据进行对比,结果表明识别误差在4.00 d内,平均误差为2.69 d。
Using the perceptual hash algorithm,the characteristics of different growth stages of rice were compared,and the image similarity was compared to realize the automatic recognition of rice growth stages.The hash code of each image to be tested was extracted by the perceptual hash algorithm,and compared with the rice image hash code in the image database that had been marked for growth stages,the label represented by the hash code with the highest degree of recognition was found,and the identification was recognized.Eight major growth stages of transplanting,regreening,tillering,jointing,booting,heading,milking and maturity stages of rice were detected,and experiments on robustness,sensitivity and retrieval accuracy for the algorithm were carried out.Comparing the results of the automatic identification with the artificial observations of the meteorological observatory,the results showed that the recognition error was within 4.00 days,and the average error was 2.69 days.
作者
张涵笑
吴苏
王苗苗
谢婷婷
ZHANG Han-xiao;WU Su;WANG Miao-miao(Henan Zhongyuan Optical-electric Measurement and Control Technology Co.,Ltd.,Zhengzhou,Henan 450000)
出处
《安徽农业科学》
CAS
2020年第10期205-207,212,共4页
Journal of Anhui Agricultural Sciences
关键词
水稻
感知哈希
发育期识别
图像检索
Rice
Perceptual hash
Recognition of the crop growing stages
Image retrieval
