Technology to validate the authenticity of varieties - a case study using apples of scientific principles and potential applications

苹果品种的准确描述是一个挑战，因为许多特征与果实的细微特征有关，而且到目前为止还不清楚哪种特征是提供最佳描述的最小字符集。今年在雷丁进行的一个本科生项目，利用了布罗格代尔国家水果收藏中心的苹果种质，在识别关键形态特征集的目标上取得了重大进展。最有趣的是，计算机辅助形状分析表明，当适当量化时，苹果形状似乎是一个特别强大的特征。该项目还表明，颜色以前只用于定性，现在也可以使用比色法进行量化，并提供了相当大的潜力作为字符。与描述相联系的是身份识别。无论是业余爱好者还是商业人士，人们对快速有效地鉴定苹果都很感兴趣。例子很多，从辨认“花园底部的那个苹果”，到辨别(超市货架上出售的)不是它们声称的苹果这一具有商业意义的问题。然而，世界上有数以千计的苹果品种：英国国家苹果登记处列出了大约1000个品种。据了解，1853至1968年间，英国种植了6000个苹果品种；布罗格代尔的国家水果收藏中心目前保存着2000多个苹果品种。客观鉴定非常困难，只有少数有多年经验的专家才能根据外观命名品种。基于DNA的鉴定昂贵、缓慢且具有破坏性。计算机化的图像分析提供了将数千种可能性减少到几十种可能性的可能性，其中人工识别对于经验较少的人是可行的。与其他一些(容易测量的)字符相结合，可能性的数量可能会进一步减少，直到可能只存在品种差异，例如与口味有关的差异，或颜色图案的微妙之处。我们的博士项目将扩展已经进行的初步分析，以包括更广泛的苹果品种；它将开发计算机辅助形状分析，并建立用于鉴定的最终特征集。它将对比使用形态学方法和DNA标记方法(微卫星和多样性阵列技术，DART)获得的结果。除了这项科学工作外，我们还将开发一个基于网络的图像分析系统，以期提供在线识别服务，需要输入少量关键字符测量数据，包括未知苹果的适当照片图像。其目的将是提供一个初步的识别、相关的正确概率和可能的替代方案。学生还将与塞恩斯伯里的合作，探索该方法是否适用于开发一种识别系统，以允许自动识别特定品种水果品系中的不同类型。还将探索形状分析在其他生鲜农产品中的潜在更广泛的应用。

The accurate description of apple cultivars is a challenge because many of the characters relate to subtle features of the fruit, and it has hitherto been unclear which is the minimum set of characters to provide an optimal description. An undergraduate project carried out this year at Reading, utilizing apple accessions from the National Fruit Collections at Brogdale, has made significant progress towards the objective of identifying a key morphological character set. Most interestingly, computer-aided shape analysis has indicated that apple shape appears to be a particularly powerful character when appropriately quantified. The project also showed that colour, previously used only qualitatively, can also be quantified using colourimetry and offers considerable potential as a character. Allied to description is identification. There is much interest, both amateur and commercial, in rapid and effective identification of apples. Examples range from the identification of 'that apple at the bottom of the garden', to the commercially significant problem of identifying apples (for sale on supermarket shelves) which are not what they claim to be. There are however, many thousands of apple cultivars in the world: the National Apple Register of the UK lists approx. 6,000 apple cultivars known to have been grown in the UK between 1853 and 1968; and there are over 2,000 cultivars of apple currently held in the National Fruit Collections at Brogdale. Objective identification is very difficult and only a few experts with many years experience can name cultivars by their appearance. DNA-based identification is expensive, slow and destructive. Computerised image analysis offers the potential to reduce thousands of possibilities down to a few tens of possibilities where manual identification is practical for those with less experience. In combination with a few other (easily-measured) characters the number of possibilities could be reduced still further until only cultivar differences relating, for example, to taste, or subtleties in colour patterning are likely to remain. Our PhD project would extend the preliminary analysis already carried out to include a much wider range of apple cultivars; it would develop the computer-aided analysis of shape and establish a definitive character set for identification. It would contrast the results obtained using a morphological approach with DNA marker methods (microsatellites and Diversity Array Technology, DArT). Alongside this scientific work, we would develop an image-analysis Web-based system with a view to providing an on-line identification service, requiring the input of a small number of key character measurements, including an appropriate photographic image of the unknown apples. The aim would be to provide a preliminary identification, associated probability of correctness, and possible alternatives. The student would also, in conjunction with Sainsbury's, explore the suitability of the approach for development of an identification system to allow automated identification of off-types in variety-specific lines of fruit. The potential wider applications of shape analysis to other fresh produce would also be explored.