基于Gabor滤波和类内PCA的昆虫识别

摘要

This paper proposes an insect recognition method based on Gabor filter and with-in class PCA in view of the low insect recognition rate. First, the pre-processed insect images are conducted Gabor filtration according to their category and extracts Gabor feature, then PCA is used to reduce the dimensionality followed by extracting the eigenvectors, at last k nearest neighbour classification is adopted to recognise the insects. Experiments show that this method has a high recognition rate, according to the test on 15 kinds of insects collection, the recognition rate is up to 95% , better than the insect recognition methods based on PCA and on Gabor filter plus PCA.%针对昆虫识别率不高的问题,提出一种基于Gabor滤波和类内PCA的昆虫识别方法.首先对经过预处理的昆虫图片按类别进行Gabor滤波,提取Gabor特征,然后利用PCA进行降维,提取特征向量,最后采用k近邻分类方法识别昆虫.实验表明,该方法识别率较高,在15种昆虫图片集上进行测试,识别率达到95％,优于基于PCA的识别方法、基于Gabor滤波和PCA的昆虫识别方法.