【作者】 刘敏娟； 洪添胜； 李震； 吴伟斌； 刘志壮；

【Author】 Liu Minjuan1,2,Hong Tiansheng1,Li Zhen1,Wu Weibin1,Liu Zhizhuang1 (College of Engineering,South China Agricultural University, Guangzhou 510642,China)

【机构】 华南农业大学南方农业机械与装备关键技术省部共建教育部重点实验室； 华南农业大学工程学院； 中国农业科学院农业信息研究所；

【摘要】 为提高荔枝商品化处理水平,搭建了一套适合荔枝等水果进行品质自动检测与分级的计算机视觉系统,通过对图像灰度变换、去除噪声、图像增强、图像分割等多种低层处理方法的对比研究,确定了适合于荔枝检测与分级的预处理方法。试验结果表明,利用荔枝横径测量的最大轴旋转法测量并分级的桂味和糯米糍两种荔枝正确率分别为98.4%和96.7%;利用荔枝最大横径处面积估测荔枝果重,该方法估测的荔枝重量与其最大横径处面积的相关系数分别可达0.9841和0.9775,分级精度可达96.8%和96.7%。更多还原

【Abstract】 The inspection of litchi’s characteristics and classification are the principal focus of the research, which aims at increasing precision of detection, improving the ways of classification and quality value work of food products in our country. A set of computer vision hardware system which is use for quality evaluation and classification of litchi was set up. Mathematical models, which transfer the process of image to operation of matrix, were built to describe litchi images. Comparing different ways of lower-layer image processing technique, a right method to satisfy the demand of specific detection and pattern recognition was determined. After studying the traditional methods of measuring diameter, this research bring forward a new way to measure diameter more accurate and faster. Its classification precision reached 98.4% and 96.7%. An equation which reflects the relationship between the area in max diameter and weight of litchi was set up to measure the weight of litchi. Its correlative coefficient was 0.9841 and 0.9775. Its classification precision reached 96.7% and 96.8%.更多还原