基于多光谱成像技术的水稻叶瘟检测分级方法研究

实时、可靠的植物病害检测是进行科学的植物喷药作业的基础,也是精确农作的关键技术之一.目前水稻稻瘟病害检测鉴定方法存在着专业知识要求高、花费大、效率低等缺点.文章提出了利用包含绿、红、近红外三波段通道的多光谱成像技术对水稻叶瘟病进行检测.研究日的是建立能够快速、准确分析稻叶瘟病情的检测模型,实时过滤掉背景噪声、自然枯叶等干扰因素,实现对水稻生长状况进行及时、有效、非破坏性榆测.研究表明,利用多光谱成像技术提取水稻叶面及冠层图像信息,可以快速有效地检测稻叶瘟病情.通过实验建立的稻叶瘟病情检测分级模型,对于营养生长期的水稻苗瘟的识别准确率为98%,叶瘟的识别准确率为90%,为实施科学的稻叶瘟防治提供了决策支持.

Identification and Classification of Rice Leaf Blast Based on Multi-Spectral Imaging Sensor

Site-specific variable pesticide application is one of the major precision crop production management operations. Rice blast is a severe threat for rice production. Traditional chemistry methods can do the accurate crop disease identification, however they are time-consuming, require being executed by professionals and are of high cost. Crop disease identification and classification by human sight need special crop protection knowledge, and is low efficient. To obtain fast, reliable, accurate rice blast disease information is essential for achieving effective site-specific pesticide applications and crop management. The present paper describes a multi-spectral leaf blast identification and classification image sensor, which uses three channels of crop leaf and canopy images. The objective of this work was to develop and evaluate an algorithm under simplified lighting conditions for identifying damaged rice plants by the leaf blast using digital color images. Based on the results obtained from this study, the seed blast identification accuracy can be achieved at 95 %, and the leaf blast identification accuracy can be achieved at 90% during the rice growing season. Thus it can be concluded that mult-spectral camera can provide sufficient information to perform reasonable rice leaf blast estimation.