徐 愫1,2，邢克智1,2，田云臣2,3，马国强3，包 蕊4，路 宁4
（1天津农学院水产学院，天津300384；
2 天津市水产生态与养殖重点实验室，天津300384；
3天津农学院计算机与信息工程学院，天津300384；
4 天津市农村工作委员会信息中心，天津 300061）

摘要：应用机器视觉技术对点带石斑鱼异常行为进行识别并及时报警有助于提高点带石斑鱼存活率，减少养殖过程中人力损耗。在养殖水体中氨氮、温度、pH等不变的条件下，通过人工调节水体溶氧浓度以获取点带石斑鱼正常与异常状态下的图像，对采集到的图像使用前景提取、二值化、开运算、中值滤波等方法进行处理后，选取正常状态与异常状态下两张图像进行差分运算，以得到两种状态下鱼口面积差图。利用鱼眼形状获取鱼眼中心坐标，由此剪裁出只包含鱼口的目标图像。根据鱼口面积判断鱼口张、闭状态与时长，若鱼口持续张开2 min，视为鱼类出现异常行为并进行报警。利用50条大小相近的点带石斑鱼对实验方法进行验证，结果表明，实验方法能有效识别点带石斑鱼的异常行为。该方法的提出有利于提高水产养殖效益，促进水产养殖现代化发展。

关键词：点带石斑鱼；图像处理；异常行为；机器视觉；报警

Study on the identification of abnormal behavior of Epinephelus malabaricus based on machine vision
XU Su1，XING Kezhi1,2，TIAN Yunchen3，MA Guoqiang3，BAO Rui4，LU Ning4
(1 College of Aquariculture, Tianjin gricultura University , Tanjin 300384, China
2 Tianjin Key Laboratory of aqua-Ecology and aquaculture, Tanjin 300384, China
3 College of computer and Information Engineering, Tianjin Agricultura University, Tanjin 300384, China
4 Information center of Tianjin Rural Work Committee, Tanjin 300000, China)

Abstract: The application of machine vision technology on Epinephelus malabaricus’ abnormal behavior identification and alarm in time is helpful to improve the Epinephelus malabaricus’ survival rate, reduce the loss of human resource in the breeding process. Under the condition of constant ammonia nitrogen, temperature, Ph value, through artificial adjusted water dissolved oxygen concentration in order to get Epinephelus malabaricus’ images under normal and abnormal conditions, using foreground extraction, binarization, open operation, median filtering method to process collected images , then two images undered normal condition and abnormal condition were selected for differential operation to get the differential Image. Using the fisheye shape to get the fisheye center coordinates, Cuting out the image contained only the fish mouth by the fisheye center coordinates. According to the fish mouth area to judge the fish mouth is closed or opened and how long is the fish mouth closed or opened, If the fish mouth is opening sustained for 2 minutes, as the fish appeared abnormal behavior and alarm. Using 50 Epinephelus malabaricus of equal size to verify this method, indicated that the method can effectively identify the abnormal behavior of Epinephelus malabaricus. The proposed method is beneficial to improve the efficiency of the aquaculture, promote the development of modernization of aquaculture.

Key words: Epinephelus malabaricus; image processing; abnormal behavior; machine vision; alarm