基于多光谱图像的水稻穗颈瘟严重度识别研究

Measurement of Rice Neck Blasts Severity by Using Multi-spectral Imagine

为实时获取作物病害程度信息,研究了水稻穗颈瘟的多图像分割方法,并利用分割结果来估测穗颈瘟的发病程度。由于图像中背景复杂,从获取的多光谱图像中提取IR、R、G分量,难以将水稻穗颈从背景中分割出来。采用IR-G和2IR-R-G分割方法将水稻穗颈从图像中分离出来,31幅图像中水稻穗颈灰度值与病情指数呈极显著线性相关,由此建立了水稻穗颈瘟严重度的多光谱图像预测模型,实现了对水稻水稻穗颈瘟的快速、准确、非破坏性检测。

In order to instantly acquire the information of disease degree of plant diseases and pests, a method of segmentation of multi-spectral image of rice neck blasts was proposed. The disease degree of rice neck blast could be calculated by the results of image segmentation. It was difficult to segment rice neck blast based on IR, R, and G imagines from complicated background by conventional calculating ways. Then IR-G and 2IR-R-G images were used to segment the rice neck blast from background. The result showed that there was high significant relationship between the disease index of rice neck blast and the gray values of IR-G and 2IR-R-G images. Based on this result, it implied that rice neck blast severity could be measured rapidly, truly and non-destructive by using a muhi-spectral CCD camera.