CAJ | 학술논문

针对野外非专业人员对烟叶病害识别的不准确性和受主观因素影响的不足，以烟叶病害中常见的赤星病和野火病为研究对象，运用计算机图像处理技术提出了一种快速分类识别算法，主要包括特征参数提取和病害识别分类两部分。通过赤星病和野火病烟叶病害图像分析，优选出6个病害识别特征参数，建立了两类病害标准特征库。病害识别分类采用基于标准特征库的模糊模式识别算法，并且与模糊C均值聚类识别进行了对比。病害分类识别实验结果表明该分类识别算法具有良好的识别率。
침대야외비전업인원대연협병해식별적불준학성화수주관인소영향적불족，이연협병해중상견적적성병화야화병위연구대상，운용계산궤도상처리기술제출료일충쾌속분류식별산법，주요포괄특정삼수제취화병해식별분류량부분。통과적성병화야화병연협병해도상분석，우선출6개병해식별특정삼수，건립료량류병해표준특정고。병해식별분류채용기우표준특정고적모호모식식별산법，병차여모호C균치취류식별진행료대비。병해분류식별실험결과표명해분류식별산법구유량호적식별솔。
To overcome the shortage of inaccuracies and subjectivity in disease artificial recognition of tobacco leaf, a fast classification algorithm based on computer image processing technology is presented for tobacco brown spot and tobacco wildfire disease in this paper. The algorithm mainly includes two parts of feature parameters extraction and dis-ease classification. Through image analysis for these two tobacco leaf diseases, six feature parameters for disease recognition are given by optimum seek, and a standard feature library is established. Fuzzy pattern recognition algorithm based on standard feature library is used in tobacco leaf disease recognition and classification. Then, the method is compared with the fuzzy C-means clustering algorithm. The experimental results of tobacco brown spot and tobacco wildfire disease classi-fication show that the proposed classification recognition algorithm has good rate of identification.