基于鱼体背部弯曲潜能算法的四种主养鱼类识别

以四种主养淡水鱼鲫(Carassius auratus)、草鱼(Ctenopharyngodon idellus)、团头鲂(Megalobrama amblycephala)、鲤(Cyprinus carpio)为例,提出一种基于图像处理技术的鱼种类自动识别方法。首先通过鱼体信息采集系统获取待识别鱼体图像,并对其进行灰度化与二值化,得到鱼体轮廓信息;然后采用邻域边界算法对鱼体的轮廓进行提取,根据轮廓曲线建立鱼体背部轮廓数学模型;最后根据轮廓模型,采用鱼体背部弯曲潜能算法对不同种类鱼体样本的背部弯曲潜能值进行计算和聚类统计,得到不同鱼类样本的背部弯曲潜能值分布区间,从而通过比对待识别鱼体背部弯曲潜能值的区间实现对不同鱼类的自动识别。对四种主养鱼类的实验测试结果表明,对团头鲂的识别精度可以达到100%,对鲫、鲤和草鱼的识别精度达到96%,基本上能准确实现四种鱼体的分类识别,具有较好的实际应用价值。

Identification of four fish species based on bending potential algorithm of fish back contour

In order to realize the automatic recognition method for different fish species such as the crucian carp (Carassius auratus), grass carp(Ctenopharyngodon idellus), megalobrama amblycephala and common carp(Cyprinus carpio), a new method based on image processing technology was proposed in this paper .Firstly, the fish images were collected which used the fish information acquisition system , and the fish contour information were calculated by the grayscale and binariza -tion algorithms.Then the mathematical model of the fish back contour was established according to the contour curve which computed by neighborhood boundary algorithm .Lastly, the bending potential values of fish back were calculated to cluste -ring in different zone.It could be got the distribution range of bending potential value for different fish species , and identi-fies the fish species which according to the zone of fish back bending potential value .The experiment results were shown that this method can identify the fish effectively .The recognition accuracy of megalobrama amblycephala can reached 100%, and the crucian carp, common carp and grass carp can reached 96%.It is achieved the accurate classification of four fish species, and it has very good practical value .