应用多光谱数字图像识别苗期作物与杂草

通过对多光谱成像仪获得的数字图片,采用一定的目标分割与形态学处理,对豆苗和杂草进行识别判断.为解决识别速度与正确率的矛盾,以豆苗和杂草图像的识别为例,提出一种基于多光谱图像算法的杂草识别新方法.应用3CCD多光谱成像仪获取豆苗与杂草图像,以多光谱图像的近红外IR通道图像为基础,利用图像分割和形态学方法,将所有豆苗叶子影像提取出来.对于剩下的2种杂草(牛筋草,空心莲子草)图像,先利用图像分析工具统计出图像块的长度、宽度、面积等基本特征参数,并根据它们形状的不同,总结出两条简单的判别规则,进行进一步的识别.本试验对147个目标进行判断,其中误判14个,正确率为90.5%,表明该方法算法简单、计算量小、速度快,能够有效识别这2种杂草,为田间杂草的快速识别提供了一种新方法.

Identification of crop and weed in seedling stage based on multi-spectral images

As machine vision owns many merits,3CCD multi-spectral camera was used to capture pictures of soybean and two common weeds(Alternanthera philoxeroides and Eleusine indica).Based on the image from IR channel of multi-spectral images and the usage of segmentation and morphological processing,the soybean image could be extracted.Although the two weeds were similar in size and color,they differed from each other by shape.So,parameters like length,width,areas of the images were calculated and two simple rules were used to identify them.In the experiment of 147 goals,14 were misjudged,and the correct recognition rate of the two weeds was 90.5%.So,this method can be used to identify these two specific weeds from each other.As the soybean leaf size was much larger than that of weeds,it could be totally extracted and identified.The test results show that the new method is simple and fast to identify crop and weed in seedling stage.