【作者】 陈坤杰；

【导师】 姬长英；

【作者基本信息】 南京农业大学 ， 农业机械化工程， 2005， 博士

【摘要】 建立牛肉分级制度，是引导市场消费、规范牛肉生产与流通、形成优质优价体系从而促进牛肉业健康快速发展的重要措施。目前世界上主要的牛肉生产及消费国均已建立了各自的牛肉分级体系，成为其保障国内肉牛业健康发展的主要手段，我国也已经制订并颁布了我国牛肉质量评定分级的行业标准。 目前，世界各国的牛肉分级体系所采用的分级方法，主要是主观的视觉评定。评价指标包括牛肉的大理石花纹、牛肉色泽、牛肉的生理成熟度以及牛肉的脂肪色等，其中，牛肉大理石花纹等级是最主要的评价指标。由于现存的牛肉等级评定方法是一种主观、定性而非客观定量的评价方法，评定过程会受到人为因素的干扰，不仅效率低，难以保证公平和公正，而且还存在相当大的误差。因此，研究牛肉分级的定量指标和客观评定方法，探讨牛肉的自动分级技术，正成为目前世界肉类生产发达国家在肉类科学领域的热点研究问题之一。 由于牛肉的生物特质，其眼肌切面大理石花纹和色泽千变万化，要找出一个能自动对其等级进行评判的方法非常困难，直到80年代初，一些科学家将机器视觉与图像处理技术应用到牛肉自动分级中，牛肉自动分级技术才获得重要发展，目前，机器视觉和图像处理技术，被公认为是实现牛肉自动分级的最有效技术手段。 本研究利用计算机图像处理和机器视觉技术，对牛胴体眼肌切面图像进行预处理，分割出大理石花纹，然后测定牛肉大理石花纹的分形维，研究其分形特征，最后找出基于牛肉大理石花纹分形维数的、能全面、准确描述牛肉大理石花纹丰富程度的量化指标，建立该指标与牛肉大理石花纹等级关系的数学模型，实现牛肉大理石花纹等级的自动判别。 牛胴体眼肌切面图像的预处理主要包括背景消除、背长肌区域分割及肌内脂肪分割等3个部分。背景消除采用区域生长法进行，生长从图像的最左上角开始，逐渐向右向下进行。生长规则采用基于区域灰度差准则，根据区域内像素亮度的灰度值来描述区域的一致性或相似性，即：对于像素g，检查其邻域中另一个像素g′，如果二者的灰度值之差在某个预先设定的范围Δ内，则把g′合并到g所在的区域内。区域生长的中止条件由图像分割的最佳阈值决定，最佳阈值通过迭代法获得。实验结果表明，采用区域生长法进行牛胴体眼肌切面图像的背景消除，可以获得良好的分割效果。更多还原

【Abstract】 Grading of beef quality needs to be performed to ensure fair deals, smooth distribution, reasonable prices and to facilitate value-based marketing. At present, the major countries in production and consumption of beef meat have established and put their beef grading standards into practice. Although an official beef grading system has not been put into practice in China, the beef grading standards were worked out and published in 2002.Quality grades of beef are primarily based on visual appraisal of the longissimus dorsi muscle. The determinants include marbling level, meat color and brightness, firmness and texture of meat, color and luster of fat. Among these determinants, the marbling level is the dominant parameter in deciding the beef quality, which is usually determined by authorized experts called graders according to the marbling abundance. Since the grading of beef marbling is largely determined by the subjective experience of the graders, there are inconsistencies and errors in judgment Therefore, to find objective and quantitative measure of the abundance degree of beef marbling and realize automatically grading the beef marbling have been attracting considerable attention since 1980’s.Many technologies have been applied to the beef quality grading, however, no a great progress has been made until some scientists began to employ machine vision and images processing techniques in their researches on beef grading in the early 80s. Computer vision has been recognized as the most promising approach to objective assessment of beef quality.The image processing of beef rib-eye cross-section consists of three parts: background removal, isolation of the LD muscle and segmentation of marbling from the LD muscle. The region growing was used to remove the background of image of beef rib-eye cross-section. The growing started from the top and left of the image and gradually went on towards the down and the right. The growing rule was established on the basis of difference of gray level of the region. That was to say, the similarity or identity of regions was determined according to the luminosity of pixel: to a pixel g, check the g’, one of pixels in its adjacent area, if the difference of gray level between g and g’ was less than the Δ, a thresh which was given in advance, then the g’ was merged into the region the g was in. The optimum threshold decided when the region growing was stopped, while the optimum threshold was calculated by using the repeat method. The experiment results showed that good segmentation could be obtained when the region growing was utilized for elimination of background of beef rib-eye cross-section image.A comprehensive image processing technique based on morphology was employed for the isolation of the LD muscle. At first, binarize the image of beef rib-eye cross-section from which the background更多还原