采用显微图像识别技术快速检测食品细菌总数

提出利用显微图像技术与BP神经网络技术检测食品细菌总数的方法,以适应快速准确检测的要求.设计了一套基于计算机视觉技术的食品细菌总数检测系统.该系统运用显微图像技术和BP神经网络技术对试验样本的细菌总数进行分析和计数.该系统由Visual C 6.0平台开发,对细菌图像的分析快速、准确,一帧图像可采集的细菌数量小于500个,对1组图像(10帧图像)的分析时间不大于30 s.比较平板计数法,该方法在牛奶、果汁和牛肉的细菌总数检测试验中,两种方法结果无显著差异(t检验结果p＞0.05).

Rapid detection of total number of bacteria in food using digital micro-image identification technique

A detection system for recognizing the total number of bacteria in food was developed. The system based on computer vision and artificial neural network technique can meet the need of detecting total number of bacterial in food rapidly and accurately. The software of the system was integratively programed by C++ language. By using BP neural network technique to analyze the microscopic image of food, the frame acquisitive counts are less than 500 and analysis-time of each group image (10 frames) is less than 30 s. Actual tests show that the results of milk juice and beef are of no significant differences between the new system and traditional method (p>0.05).