【摘要】 利用计算机视觉技术对冷却牛肉的新鲜度进行了分析研究 ,应用 BP神经网络对牛肉脂肪组织进行了分割 ,分割正确率可达 97%以上。选取图像原始颜色信息作为评价新鲜度的特征值组成特征向量 ,采用 HSI、RGB和 CMYK 3种彩色模型评价冷却牛肉的新鲜度 ,其准确率分别为 82 .1 %、78.6 %、75 .0 %。基于计算机视觉技术的冷却牛肉新鲜度评价方法与实验室分析方法相比 ,评价准确率有明显提高 ,并且没有繁琐实验过程 ,是一种更高效的牛肉新鲜度评价方法更多还原

【Abstract】 In this paper, a computer vision technique was applied to estimate freshness of chilled beef. A BP neural network was used to separate fat structure from the image of chilled beef, in which the discriminating rate was higher than 97%. The color features of muscle structure were inspected by the color models of HSI, RGB and CMYK. Two image analysis methods that take pix values of a pix and color information of whole image as inputs respectively were inspected in the research. The results showed that the effect taking the statistical features of a whole image as input was better than that taking the color pix values of a pix. The appraising accuracies of three color models of HSI, RGB, CMYK are 82 1%, 78 6%, 75 0% respectively. The research results also showed that the accuracy of the appraising method of freshness of chilled beef based on computer vision technique was evidently higher than that of biochemistry methods in laboratory.更多还原