摘要
传统的昆虫识别方法费时费力,应用图像处理技术提取昆虫图像视觉特征,实现昆虫机器自动识别,可以解决传统方法的不足.本研究依据纹理是昆虫分类的重要特征,应用角度无关的Gabor滤波器提取昆虫图像的纹理特征,然后用SVM算法分类,实验结果表明:角度无关Gabor-SVM昆虫识别方法正确率为80%,是比传统Gabor和灰值游程矩阵更好的识别算法,该方法能较准确识别昆虫,省时省力.
The traditional method of recognizing insects is time-consuming. Its solution is using image processing technology to extract visual features of insect images and then machines automatically identify insects. According to texture is important feature for insects recognition, in our research, angle invariant Gabor is used to extract texture feature of insects images, and then use SVM algorithm to classify insects. The experimental results show that：the correct rate of angle invariant Gabor-SVM algorithm for insect recognition is 80%, which is better than traditional Gabor and run length algorithm, invariant Gabor-SVM algorithm can more accurately identify insects and time-saving.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第1期143-146,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金重点项目(60736008)资助
国家“八六三”高技术研究发(2008AA01Z301)资助
国家自然科学基金项目(60873094)资助
