An image segmentation method in insect recognition was developed. The insect separating from the background image is a key step in insect automatic recognition. Taking the Zhejiang 66 major quarantine pests as the reserach objects,this paper uses Otsu algorithm to segment insect image. It uses gray-scale process to accelerate the speed of segmentation,and then uses median filter to remove most of the noise, and finally uses the Otsu algorithm to segment insect image. The paper uses mathematical morphology to amend the shortcomings. Mathematical morphology methods such as dilation, erosion are applied to image enhancement and then Otsu algorithm is used to segment the image. The experimental results domenstrate the effectiveness of this method.%以浙江省66种重大检疫性害虫为研究对象,采用Otsu算法实现昆虫图像的分割,取得了较好的效果.首先对昆虫图像进行灰度化处理,加快分割速度,再对灰度图像进行中值滤波,去除大部分噪声,最后,用Otsu算法对灰度化和中值滤波后的昆虫图像进行分割.针对存在的不足,运用数学形态学方法进行修正.首先利用数学形态学中的膨胀和腐蚀运算对昆虫图像进行增强,然后运用Otsu算法对图像进行分割.实验结果表明,将Otsu算法与数学形态学方法结合起来能够有效地提取昆虫图像.
展开▼
