High-speed weight estimation of whole herring (Clupea harengus) using 3D machine vision

Abstract: Weight is an important parameter by which the price of whole herring (Clupea harengus) is determined. Current mechanical weight graders are capable of a high throughput but have a relatively low accuracy. For this reason, there is a need for a more accurate high‐speed weight estimation of whole herring. A 3‐dimensional (3D) machine vision system was developed for high‐speed weight estimation of whole herring. The system uses a 3D laser triangulation system above a conveyor belt moving at a speed of 1000 mm/s. Weight prediction models were developed for several feature sets, and a linear regression model using several 2‐dimensional (2D) and 3D features enabled more accurate weight estimation than using 3D volume only. Using the combined 2D and 3D features, the root mean square error of cross‐validation was 5.6 g, and the worst‐case prediction error, evaluated by cross‐validation, was ±14 g, for a sample (n = 179) of fresh whole herring. The proposed system has the potential to enable high‐speed and accurate weight estimation of whole herring in the processing plants. Practical Application: The 3D machine vision system presented in this article enables high‐speed and accurate weight estimation of whole herring, thus enabling an increase in profitability for the pelagic primary processors through a more accurate weight grading.     

基于可见/近红外高光谱成像技术的鸡蛋新鲜度无损检测

以400~1 000nm高光谱系统获得鸡蛋样本的高光谱图像,利用蒙特卡洛法检测异常样本,采用不同预处理方法处理原始光谱;应用竞争性正自适应加权算法(Competitive Adaptive Reweighted Sampling,CARS)、遗传偏最小二乘法(Genetic Algorithms PLS,GAPLS)和间隔蛙跳法(Interval Random Frog,IRF)对预处理后光谱数据提取特征波长;分别建立基于全光谱和特征波长的偏最小二乘回归(Partial Least Squares Regression,PLSR)和最小二乘支持向量机(Least Squares Support Vector Machine,LS-SVM)鸡蛋新鲜度预测模型。结果表明:标准正态变量变换(Standardized Normal Variate,SNV)法为最优预处理方法;利用CARS、GAPLS和IRF分别选出8,35,74个特征波长;基于GAPLS提取的特征波长的LS-SVM模型最优,其校正相关系数(Rc)为0.899,预测相关系数(Rp)为0.832。表明基于高光谱成像技术的鸡蛋新鲜度无损检测是可行的。     

基于机器视觉的鸡蛋品质无损检测方法

目的以蛋品质量评价指标—哈夫单位值为检测标准,建立一种基于机器视觉的鸡蛋品质无损检测方法。方法通过摄像头捕捉到鸡蛋图像信息,利用MATLAB对鸡蛋图像的G分量以及I分量进行特征参数提取,并计算出与鸡蛋新鲜度相关的4个特征参数:蛋黄面积比、气室面积比、气室高度比与气室直径比,将其作为自变量,通过高精度游标卡尺实测每个鸡蛋样本哈夫值作为因变量,分别建立一元回归模型,寻找特征参数与哈夫值的关系,并根据哈夫值对鸡蛋新鲜度进行分级。结果实验表明,所测4个特征参数中,蛋黄面积比与哈夫值存在较强的相关性,相关系数为0.78,拟合优度为0.62,蛋黄面积比越小,鸡蛋哈夫值越大,说明鸡蛋越新鲜。结论基于机器视觉的鸡蛋品质无损检测方法不仅具有较强的应用价值,还可以为鸡蛋品质智能分级提供技术支撑。     

基于高光谱图像与光谱特征融合技术的鸡蛋新鲜度无损判别模型的建立

鸡蛋新鲜度等级评价是鸡蛋品质检测过程中的一项重要技术指标。选取了不同储藏环境的鸡蛋样本并采集其高光谱图像信息与光谱信息,提取图像特征和光谱特征;采用并行式融合方法进行图谱特征融合,基于连续投影法-灰度共生矩阵方法进行特征提取;建立支持向量机鸡蛋新鲜度判别模型。采用粒子群算法优化模型,训练集准确率达到85%,预测集准确率达到76.67%。为了解决单模型可能出现的偶然性误判问题,采用递进式特征融合方法,引入多模型共识策略和深度残差网络ResNet 50分析方法。建立基于连续投影法-方向梯度直方图特征提取方法的多模型共识策略,该模型的训练集准确率提升至89%,预测集准确率提升至88%;同时,建立基于连续投影法-方向梯度直方图特征提取方法的深度残差网络ResNet 50模型,模型的训练集准确率提升至89%,预测集的准确率提升至86.67%。图谱特征融合建模分析表明,并行式融合方法和递进式融合方法对鸡蛋新鲜度等级判别都有一定的可识别性,且递进式融合算法的多模型共识策略判别效果更佳。     

鸡蛋新鲜度指标与贮藏天数相关性研究

目的 分析鸡蛋新鲜度指标与贮藏天数的相关性, 确定新鲜度指标数值显著变化的时间, 建立新鲜度与贮藏时间相关性模型。方法 2 d为一个时间点, 设立12个实验组, 每组5枚鸡蛋, 按照标准中鸡蛋新鲜度评定指标重复3次测定鸡蛋相对密度、气室高度、哈夫单位、蛋黄指数、pH值、失重率。根据Pearson相关系数评价贮藏天数与鸡蛋新鲜度各指标之间的相关程度, Duncan法评价贮藏天数与鸡蛋新鲜度各指标之间的差异显著程度。结果 鸡蛋新鲜度各指标与贮藏天数之间显著相关(P<0.05), 各新鲜度指标数值出现显著变化的时间各不相同, 各天数之间的鸡蛋失重量、失重率差异显著(P<0.05)。结论 本研究所建立的鸡蛋新鲜度与贮藏时间的相关性模型, 可作为鸡蛋新鲜度评价模型。     

An indirect approach for egg weight sorting using image processing

An indirect approach for egg weight sorting, using image processing, is proposed in the paper. The eggs are sorted in four classes by weight. Regression analysis is used for approximation of relationship between egg weight and egg geometric parameters—perimeter, area, major and minor axis, shape index and shape factor. The values of the geometric parameters, collected by image processing and the one, collected by traditional method, are compared for each egg sample, using percent differences between data. The experimental results show that the most significant parameter for egg weight indirect measurement is the egg area, with correlation coefficient 0.989. The mathematical model for the relationship between weight and area of the egg is defined with coefficient of determination 0.978. The classification accuracy is achieved within the eggs test sample sorting. The total classification error is 2.5% for test set and 12.5% for training set.     

机器视觉在蚕茧表面积测量中的应用

针对传统评茧方法的缺陷,提出了基于机器视觉技术的蚕茧表面积测量方法。利用机器视觉系统拍摄蚕茧图像,根据图像的灰度分布特征,通过Otsu法分割蚕茧图像得到二值化图像,将蚕茧看作椭圆体,由蚕茧区域的长轴和短轴值计算出蚕茧表面积。根据蚕茧的边缘图像,利用最小矩形法确定椭圆长轴和短轴的值,可以测量任意旋转角度的蚕茧表面积。实验结果表明该方法是有效的,为科学评茧提供了依据。     

On-line estimation of eggshell strength based on acoustic impulse response analysis

This paper proposes a nondestructive method for on-line estimation of eggshell strength based on acoustic resonance analysis. The system employed digital signal processing (DSP, TMS320F2812) as core processor to collect and analyze the response signal of eggshell. Fast Fourier Transform (FFT) and Discrete Cosine Transform (DCT) were used to transform the time domain signal into frequency domain signal for next analysis. Synergy interval partial least square (si-PLS) combined with multiple stepwise regression was used to establish a linear calibration model for eggshell strength measurement. The performance of the optimal model using 11 frequency variables was achieved, with R (correlation coefficient) of 0.776 and RMSEP (root mean square error of prediction) of 3.010 in prediction set. Good consistence confirmed that the acoustic resonance system has significant potential in on-line estimation of the eggshell strength.Industrial relevanceWith processing and handling procedure continues increasing in egg production, the possibility of the presence of eggshell crack sharply increases. In such case, it is essential to measure the strength of eggshell, so as to maintain the balance between eggshell strength and the handling load in the processing of egg collection, sorting and transportation. This work proposes a nondestructive method for on-line estimation of eggshell strength based on acoustic resonance analysis, and builds a robust calibration model to improve the prediction ability. The research data presents a potential way for on-line and non-destructive measurement of eggshell strength in egg industry.     

基于IGS-SVM模型的牛肉生理成熟度预测方法

生理成熟度是判定牛肉质量等级的重要指标,本实验建立一种通过改进的网格搜索（improved grid search,IGS）算法优化支持向量机（support vector machine,SVM）参数的模型,以实现牛肉的生理成熟度的预测。收集18、36、54、72 月龄的牛肉样本各25 个,共计100 个。利用机器视觉,采集样本的显微图像,经过图像处理后,提取不同生理成熟度牛肉的肌纤维特征参数,用统计学方法分析牛肉生理成熟度和肌纤维特征参数之间的相关性,并以肌纤维特征参数作为输入,利用76 个训练集样本,建立牛肉生理成熟度的SVM预测模型。为优化所建立的SVM模型,提出一种IGS算法,用其对SVM模型的约束参数C以及核函数参数g进行优化,结合留一交叉验证法得到最优的（C,g）参数组合,并将最佳参数代入分类器,得到优化的牛肉生理成熟度预测模型。用24 个测试集的独立样本检测模型的适用性并估测性能,结果表明：利用该模型对牛肉生理成熟度预测的准确率可达到91.67%;与传统网格搜索算法相比,IGS算法使得模型在训练时间上缩短了1 755.41 s。这表明所建立的模型具有较好的预测效果,也说明根据牛肉肌纤维的特征参数结合机器视觉及图像分析技术,对牛肉生理成熟度进行自动判定的方法是可行的。     

Assessment of egg freshness by prediction of Haugh unit and albumen pH using an artificial neural network

Eggs are a good source of high quality protein and knowing their quality (physical and chemical properties) during storage is of great importance. Thus, the aim of this research was to design a computer vision system to assess egg freshness during storage time. To this end, 210 intact eggs were collected and stored for 30 days under room conditions (25?±?2 °C and 20?±?3%). After imaging, every other day, some internal and external quality characteristics including yolk height, yolk and albumen pH, yolk and albumen density and Haugh unit (HU) were measured as destructive parameters and area index (D) egg weight as non-destructive parameters. Based on Pearson correlation coefficients, area index were significantly correlated with all destructive variables (p?<?0.05). In order to predict egg freshness, artificial neural network was trained by Levenberg–Marquardt, scaled conjugate gradient, Bayesian regulation, resilient and radial basis algorithms. The best result of artificial neural network for HU and albumen pH prediction was achieved by the Levenberg–Marquardt algorithm with the correlation coefficient of 0.93 and 0.87, respectively.     

融合图谱特征信息的明虾挥发性盐基氮含量无损检测

为实现明虾中挥发性盐基氮（total volatile basic nitrogen,TVB-N）含量的快速预测,采用近红外光谱和机器视觉技术获取明虾图谱特征信息,融合图谱特征信息构建预测明虾中TVB-N含量的支持向量机模型。获取明虾4 ℃贮藏0～12 d共51 个样品的光谱信息和图像信息,同时参照GB 5009.228—2016《食品中TVB-N的测定》方法测定其TVB-N含量。结果表明,利用350～1 000 nm和940～1 650 nm双波段融合的光谱特征信息,并对其进行一阶导数的预处理,同时采用竞争性自适应加权算法挑选特征波长后建立的模型效果较好,其预测集相关系数（correlation coefficient in the prediction set,Rp）为0.968 7,验证集标准分析误差（standard error of prediction,SEP）为10.56 mg/100 g,相对分析误差（relative percent deviation,RPD）为3.38;利用图像特征信息所构建的模型效果较差,Rp为0.933 5,SEP为19.79 mg/100 g,RPD为1.74。最后,融合特征图谱信息构建TVB-N含量的预测模型,相比其他2 种方法,该模型精度和稳定性都得到了提高,其Rp为0.988 4,SEP为7.51 mg/100 g,RPD为6.29。该结果证实近红外光谱技术结合机器视觉方法预测明虾中TVB-N含量的潜力,为分析评价明虾在冷藏过程中新鲜度的变化规律提供了快速检测技术。     

数字图像处理技术结合偏最小二乘法定量分析柑橘质量

目的：构建柑橘质量模型,实现其计算机视觉法质量检测与在线分级。方法：以湖南湘中成熟中期涟红温州蜜柑为研究对象,对计算机视觉系统采集的各柑橘的4 幅图像进行图像裁切、图像去背景、图像二值化与取反操作后,提取柑橘区域像素,以柑橘区域像素与图像像素比为自变量,建立柑橘质量偏最小二乘回归(PLS)预测模型。结果：柑橘质量预测绝对误差为－ 14.9092～4.9981g,平均误差为－ 3.9278g,误差标准差4.5210g,± 10g 质量内的正确识别率为93.3333%,± 8g 质量内正确识别率76.6667%。结论：通过计算机视觉技术进行柑橘质量在线分级和生长期中挂果质量的监测是可行的。     

机器视觉和电子鼻融合的番茄成熟度检测方法

目的:探索机器视觉与电子鼻融合法在果蔬成熟度检测中的应用能力,实现番茄不同成熟度的检测.方法:基于机器视觉和电子鼻采集系统,提出了一种基于多源信息融合的番茄成熟度检测方法.以机器视觉筛选出的6个颜色特征和电子鼻筛选出的10个气味特征为基础,建立番茄成熟度检测的最小二乘支持向量机模型.通过试验对融合方法和单一方法进行对比...     

Rapid Assessment of Tomato Ripeness Using Visible/Near-Infrared Spectroscopy and Machine Vision

A non-destructive assessment using visible/near-infrared spectroscopy and machine vision has been investigated for measuring tomato ripeness. Relationship between spectral wavelengths and green grayscale value was evaluated by application of chemometrics techniques based on partial least squares (PLS) regression. The tomatoes were divided randomly into two groups: 170 fruits for calibration and 71 for prediction. An accurate estimation, measured with a correlation coefficient of 0.992 and root mean square errors of prediction (RMSEP) of 9.92, was obtained when using the developed PLS model built with 550–750 nm spectral range. The accuracies of calibration and validation models based on data measured in this band were 90.93 and 90.05%. The prediction accuracy for 150 external independent samples was 90.67%. The results show that it is possible to realize detection standardization of tomato maturity based on only visible spectroscopy (550–750 nm) and machine vision technologies. This detection method does not depend on a visual grading or other maturity indices as a reference. It highlights the potential of the method to determine tomato ripeness and the optimum harvest time.     

Chinese Quince (Cydonia oblonga Miller) Freshness Rapid Determination Method Using Surface Acoustic Wave Resonator Combined with Electronic Nose

In this article, Chinese quince (Cydonia oblonga Miller) freshness determination method was investigated using surface acoustic wave resonator, electronic nose, and surface acoustic wave resonator combined with electronic nose. Human sensory evaluation and weight loss index were examined as freshness reference. Results indicated that quince freshness decreased during storage procedure. Surface acoustic wave resonator output frequency and electronic nose measurement data stochastic resonance signal-to-noise ratio Eigen values characterized quince quality under different storage time. Freshness predictive models were developed using surface acoustic wave resonator frequency, electronic nose signal-to-noise ratio spectrum Eigen values, and their hybrid model. Validating experiments results demonstrated that the hybrid predictive model presented higher predicting accuracy (R² = 0.987) than other two models. The proposed method is promising in fruit quality rapid analysis.     

基于支持向量回归的核桃在线称重系统

为了对核桃重量进行在线检测,采用加速度传感器信号对称重传感器信号动态补偿校准和支持向量回归(SRV)预测方法,设计多传感器信息融合的核桃重量在线测试系统。对400枚核桃分别在速度为0.02,0.03,0.05 m/s条件下进行数据采集,并对采集的数据进行训练和校验,分析得到最优的核桃重量预测模型为基于线性核函数的SVR模型,较佳的测试速度为0.03 m/s。利用200枚核桃样本在0.03m/s的条件下进行实验验证,结果显示系统预测的核桃重量与核桃实际重量线性拟合的r~2为0.85,平均绝对误差为1.67g,表明该系统可以较为准确地实现核桃在线称重。     

Prediction of Poultry Egg Freshness Using Vis-Nir Spectroscopy with Maximum Likelihood Method

Crucial physio-chemical changes occuring in eggs during storage after laying lead to loss of egg freshness. In this research, a new method for prediction of egg freshness using transmission visible near infrared spectroscopy was investigated. For this purpose 300 eggs were stored at two control conditions: refrigerator (4–5°C, 75%RH) and room (24–25°C, 40%RH) then by special egg holder, transmission spectroscopy was measured. For two eggs groups, 25 eggs in each group, in six days were tested by spectroscopy, after that Haugh unit and air cell height was measured directly. The non-destructive visible near infrared spectroscopy spectral measurements from 300 to 1100 nm (832 length of wave) were done as well as Haugh unit, air cell height for each egg and created the database for both environments. Finally a maximum likelihood latent root regression algorithm was developed to predict Haugh unit and air cell height by spectrum observation. The database was randomly divided into two parts. Training data, was used for maximum likelihood latent root regression parameter tuning and training of the model and testing data, was used just for model evaluation. Results indicated that maximum likelihood latent root regression method showed good prediction ability with coefficient of determination (R²) value up to 0.82 and 0.86 for Haugh unit and air cell height, respectively for testing data set. The results showed this method was better in comparison with partial least square regression (R² up to 0.79 and 0.72 for air cell height and Haugh unit) which was already used for this prediction.     

基于CUVE-PLS-DA的鸡蛋新鲜度在线检测分级

针对目前鸡蛋新鲜度检测技术方法存在劳动强度大、检测精度低、分级效率不足等缺陷,本研究在4 800 枚/h的禽蛋传输机上搭建了可见-近红外透射光谱（501～1 000 nm）在线检测装置,动态采集鸡蛋透射光谱数据,并建立光谱信息与鸡蛋哈夫值等级的偏最小二乘判别模型。采用3∶1原则对鸡蛋样本进行随机划分,其中校正集169 个,验证集57 个,通过比较多种光谱预处理方法以及两种特征波长选择方法,得出标准正态变换预处理方法和多模式共识方法能够有效地提高模型的正确率、运算效率和预测能力,优化模型后的校正集和验证集准确率分别为92.31%、91.23%。结果表明本实验建立的可见-近红外光谱透射光谱检测方法能够对鸡蛋的新鲜度进行无损、智能、在线检测分级。     

联合图像最优特征提取及改进RBF神经网络的苹果质量估计

目的：以阿克苏苹果为例,设计一种联合图像最优特征提取和改进RBF神经网络学习的苹果质量估计方法,以克服人工分级称重成本高、误差大的缺陷。方法：首先,建立苹果图像采集系统,得到苹果前景图像信息;其次,设计苹果图像特征集合最佳子集提取策略,将最佳子集提取过程转化为目标函数优化问题,并利用改进的离散蝗虫优化算法进行求解,从而得到最佳苹果图像特征子集;最后,构建基于RBF神经网络学习的苹果质量估计模型,将最佳特征子集作为网络输入,并采用蝗虫优化算法优化配置RBF神经网络超参数,从而实现对苹果质量的有效估计。结果：所提苹果质量估计方法精度更高,质量估计值平均相对误差率为1.23%。结论：该方法可以有效实现苹果质量预估,也能够推广应用到其他类似轴对称形状的水果质量估计。     

在线动态称重系统开发及质量采集算法研究

以包装生产线上连续在线动态称重及后续处理的功能为导向,从称重工作站的结构、控制系统、分拣机构、数据处理与保存方面进行研究。采用PLC、A/D模块、称重仪表、称重传感器、变频传送带控制等技术,开发出一套在线动态称重系统。系统采用区间采样、区间数值平均的算法,测量和计算动态数据,使产品的动态质量值与静态值之间的差值最小。在线称重系统可根据功能快速搭建,实现对产品的质量采集、对比及对产品的分拣、分装、包装等功能。系统可快速二次开发对接包装线所需要的功能,且构造简洁实用、成本低,质量采集的准确性高,应用前景广泛。