摘要： 浮游生物图像识别分类是海洋生态研究的重要内容和必要前提，传统的浮游生物图像识别分类需要由专业人员进行人工识别，工作量大、效率低。该文提出一种基于小波分解结合颜色信息熵的浮游生物图像识别方法，提取图像的三层小波分解后系数的数学特征和四叉树分块后的颜色信息熵构造特征向量，采用相似度模型和K-近邻分类器对浮游生物图像进行分类。实验表明，与传统方法相比，该方法能在保证识别率的基础上提高识别效率，并具有良好的稳定性。

关键词: 浮游生物, 小波分解, 颜色信息熵, K-近邻

Abstract: The recognition and classification of plankton images is the important content and precondition to the research of zoology. Traditional methods of plankton images recognition is mainly to be recognized under the microscope by professional. But this method has the demerit of lower efficiency. Under this background, this paper presents a plankton image recognition based on wavelet decomposition and color information entropy. The plankton and eigenvectors are extracted from the images, and K-neighbor classification method is used to recognize the plankton images in database. Experiment proves that the recognition rate is higher than others, and this method is deemed to have high efficiency and stability.

Key words: plankton, wavelet decomposition, color information entropy, K-neighbor

中图分类号: 

• TP391.41