作　　者： ; ; ;

机构地区： 宁波大学信息科学与工程学院

出　　处： 《宁波大学学报（理工版）》 2014年第2期49-51,共3页

摘　　要： 传统梭子蟹养殖定期质量估计的方法是捕捞部分梭子蟹进行称重,会对其造成物理和应激伤害,且检测过程中受到人为因素干扰,因此利用机器视觉技术对梭子蟹的质量进行估计.通过摄像头拍摄不同生长阶段的梭子蟹图像,对图像进行模板校正及图像分割,提取梭子蟹面积特征参数.利用最小二乘法对面积和质量进行拟合,其中二次多项式相关性最好可达到0.9220,测试平均相对误差6.40%,证明该方法可以满足梭子蟹质量估计的要求. The traditional means of weight estimation for swimming crabs is catching some samples to weigh, which may raise some unwanted issues such as safety problems, high labor cost, time-consuming, human subjectivity and low efficiency. In this paper, the machine vision technology is applied to facilitate the weighing process. First, the image and the weight data of a swimming crab at different growth stages are collected. Then the image is digitally processed for correction and segmentation through which the area of swimming crab can thus be identified. Data fitting between area and weight is conducted using the least squares technique. It is found that the best fitting can be obtained using quadratic polynomial with correlation coefficient being 0.9220 and average relative being 6.40%, suggesting the validation of the proposed method for weight estimation of swimming crabs.

关 键 词： 梭子蟹 质量估计 图像处理技术 数据拟合

领　　域： [自动化与计算机技术] [自动化与计算机技术]