摘要
目前利用图象处理和模式识别技术进行浮游生物的自动分类与计数是一个研究热点,该文根据浮游蚤类图像的特点,设计了一种自动分类计数算法。该算法首先通过有效的分割手段将蚤类个体从背景区域中提取出来;然后计算与其生物学特征相关的形状和纹理特征参量;最后选择特征参量输入神经网络分类器对不同蚤类进行分类和计数。样本主要采集自胶州湾海域,图像通过zeiss显微图像采集系统获取。实验证明,由于研究目标明确,算法简单有效,取得了很好的分类计数效果,很大程度上减轻了人工分类计数的工作量。
At present it is a hotspot in the research of counting and classification of zooplankton by the image processing and pattern recognition techniques. In this paper we proposed an algorithm for automatic counting and classification which considers the characters of the zooplankton in the image. Firstly, we distinguish the individual from the background by efficient method of segmentation. Then we calculate the shape and texture features relative to the biological features of the individuals in the image, Finally, by importing the chosen feature parameters into the BP neural networks we get the result of the classification. Samples are gathered from the Jiaozhou Bay and the images are collected by the Zeiss microscopical system. Demonstrated by the experiments, accurate results can be achieved and workloads are lightened to a great extend on account of the definite research object as well as the straightforward algorithm.
出处
《计算机仿真》
CSCD
2006年第5期167-170,共4页
Computer Simulation
