基于仿生模式识别算法的海洋微藻识别研究

Recognition of marine microalgae based on bionic pattern recognition

依据微藻个体及成像的特点,给出了矩形度、能量、熵、惯性矩、相关度和局部平稳度等形状和纹理参数作为识别的特征值,并利用仿生模式识别算法对海洋微藻实现自动识别。利用文中给出的方法,对在海域中随机采集的不同形状、大小、纹理的微藻混合图像进行识别实验,结果显示,该方法能够准确识别出图像中不同种及同种不同状态下的藻体,说明该方法在微藻图像识别中是有效和可行的。

The automatic recognition of marine mic via six shape and texture parameters including rec roalgae was achieved by the bionic pattern recognition algorithm tangular degree, energy, entropy, inertia moment, correlation and local stationary degree as the characteristic values of marine microalga recognition characteristics. The microalga specimens collected randomly in marine water were experimentally recognized by the above bionic pattern recognition algorithm according to the microalgae mixed images with the differences in the shape, size or texture of the algae. The results showed that this method accurately identified the microalgae between different kinds or same kind in different states, indicating that the method is effective and feasible in microalgae image recognition.