文章摘要
张俊俊;赵 号;翟晓东;胡雪桃;邹小波;石吉勇.基于超声成像技术的方腿中异物检测[J].中国食品学报,2019,19(8):223-229
基于超声成像技术的方腿中异物检测
  
DOI:
中文关键词: 方腿  异物  超声成像检测  回波信号  最小二乘支持向量机  识别率
英文关键词: Luncheon sausage  foreign body  ultrasonic imaging  echo signal  least squares support vector machine  recognition rate
基金项目:国家重点研发计划重点专项(2017YFC1600805,2018YFD0400803);国家自然科学基金项目(31601543,31671844,31801631);江苏省自然科学基金项目(BK20160506,BK20180865)
作者单位
张俊俊;赵 号;翟晓东;胡雪桃;邹小波;石吉勇 江苏大学食品与生物工程学院江苏镇江212013 
摘要点击次数: 7
全文下载次数: 0
中文摘要:
      研究提出了方腿异物的反射超声成像检测方法,实现了异物的无损检测。首先利用扫描超声成像系统测量方腿声学参数,获得不同尺寸方腿(3 mm×3 mm、5 mm×5 mm及10 mm×10 mm)、深度(5,10,15 mm)处异物(金属、玻璃、塑料)的反射超声图像;根据方腿及异物在回波信号和超声图像中的差异,对方腿中异物进行判别分析。结果表明,空白方腿(不含异物)与含异物方腿在回波信号上差异显著,含异物方腿较无异物方腿的回波信号多一个明显回波峰,可进行异物深度检测;无异物方腿与含异物方腿的超声图像差异显著,以5 mV的声波强度为判断阈值,对所有方腿内部超声图像中的异物检测准确率达96.29%;利用平均声波反射强度、面积检出率及常见纹理特征值为特征变量,采用最小二乘支持向量机进行异物识别,当主成分数为6时,模型的识别率最高为81.67%,具有良好的识别效果。本研究提供了一种直观、简便的方腿异物检测方法,为食品中异物的检测提供了新思路。
英文摘要:
      In order to control and eliminate the harm caused by foreign bodies in luncheon sausages, a method of foreign bodies detection by ultrasonic imaging technology was proposed, and foreign bodies were detected and identified nondestructively. Ultrasonic property parameters, such as ultrasound velocity, acoustic impedance and attenuation coefficient of luncheon sausages and reflection images of different size foreign bodies buried in different depth were obtained by ultrasonic imaging system. And then, foreign bodies were detected and identified according to the reflected signal and ultrasound images. The research showed that inner acoustic images of luncheon sausage with and without foreign bodies had significant differences, the luncheon sausages contained foreign bodies generated a additional echo peak compared with controls. The ultrasound image of luncheon sausages with foreign bodies and controls had significant difference, and the detection rate of foreign bodies form 20 inner acoustic images of luncheon sausage could reach at 96.29% with a threshold of 5 mV of ultrasonic intensity. The best predictive precision of LS-SVM model based on selected characteristic variable can be reached when principal component signals were six, and precision rates of calibration set and prediction set were 93.27% and 81.67%, respectively. The result indicate that ultrasonic detecting system is a feasible way to discriminate foreign bodies in luncheon sausage quantitatively, and has a wide application prospect.
查看全文   查看/发表评论  下载PDF阅读器
关闭

手机扫一扫看
分享按钮