基于计算机视觉的大黄鱼体尺测算与体质量估测

Computer vision-based body size measurement and weight estimation of large yellow croaker

在无应激情况下获取鱼体体尺数据、质量,是鱼高产养殖的一个重要环节。这些参数作为重要的数据,将其提取分析,可大大增加养殖收成。通过采用计算机视觉的方法,提出在复杂背景下针对大黄鱼进行个体信息提取的算法。通过背景减法和去除噪声算法可去除背景干扰,可准确提取自然姿态下鱼的个体轮廓,对采集到的样本轮廓进行鱼体测点提取,识别其体测点,实现鱼体身长、全长、体高、尾鳍长的测量并完成体质量估测。经实验,鱼体体尺测量相对误差平均0.3%,利用身长与质量的关系拟合出估测体质量的关系方程。本文提出的大黄鱼的无应激体测方法为福利养殖提供了新方法。

Obtaining data and quality of fish body measurements under stress-free conditions is an important part of high-yield culture of fish. These important data extracted and analyzed which can greatly increase the breeding yield. In this paper, an algorithm for ex- tracting individual information for large yellow croaker under complex background is proposed through computer vision. Background subtraction and noise removal algorithm can remove background interference, and which can accurately extract individual contours of fish under natural posture. Contours of the collected samples were extracted from the image to find the body measurement points, and the body length, full length, body height, and caudal fin length will be measured and the body weight will be estimated. After the ex periment, the relative error of fish body size measurement is 0. 3% on average. Using the relationship between body length and weight, a relational equation for estimating body weight was fitted. The non stress measurement method for large yellow croaker is proposed in this paper and which provides a new method for welfare breeding.