中文摘要：

针对现有水果识别方法需大量水果样本学习或仅对单一特征进行识别而导致的识别率较低的问题，提出一种基于水果图像处理的水果颜色和形状特征参数的提取方法、基于灰色关联分析和模糊隶属度匹配的球形水果自动识别方法。该方法通过提取水果图像关注区域（regionofinterest，ROI）的颜色和形状特征，建立参比水果的颜色特征参比数据库和形状特征隶属度函数，计算待识别水果与参比水果颜色特征的灰色加权关联度，求取待识别水果对于参比水果形状特征参数的模糊隶属度，按各特征量等权的原则合成待识别水果对参比水果的总匹配度，并根据总匹配度的大小实现待识别水果种类的判别。大量实验结果表明：该方法简单、有效，不需要大样本量水果的学习和训练，平均识别正确率达到99％以上。

英文摘要：

Aiming at the problem that existing fruit recognition methods either need a lot of fruit samples to train or only use a single feature to do identification, which leads to low recognition rate, a feature parameter extraction method of fruit color and shape based on fruit image processing is proposed, and an automatic recognition method of spherical fruit based on grey relational analysis and fuzzy membership degree matching is proposed. The method ex tracts the fruit image color and shape features in the region of interest （ ROI）, establishes the fruit color feature refer ence database and shape feature membership functions, calculates the grey weighted relational grade of the color fea tures between the fruit to be identified and reference fruits, and calculates the fuzzy membership degree of shape fea ture parameter of the fruit to be identified relative to the reference fruits. The method synthesizes the total matching degree of the fruit to be identified relative to reference fruits according to the principle that all features are equal weight, and realizes the recognition of the fruits to be identified based on the total matching degree. Mass experiment results show that the proposed method is simple and effective; it doesn＇t need large fruit samples for learning and training, and the average recognition accuracy reaches above 99%.