Quantitative extraction and analysis of pear fruit spot phenotypes based on image recognition

The quantitative analysis of fruit spot phenotypes is useful for traits evaluation and cultivars comparison in pear (Pyrus spp.) breeding. However, it is limited by the current manual observation and measurement of fruit spot phenotypes. In this study, fruits from nine pear cultivars with large differences in fruit spots were investigated. Images of fruit epidermal samples were photographed with a fixed light source. An automatic method to measure indicators of pear fruit spots based on image recognition was proposed. Graphical interfaced software to extract and analyze pear fruit spot phenotypes was proposed. The results show that the method can automatically measure fruit spot area, amount, size, and the color difference between fruit spots and epidermis. The measurement accuracy and precision of the method is equivalent to that of manual ones and the fruit spot amount measurement is more precise. The fruit spot ratio, density and average shortest distance can also be calculated. The fruit spot index is calculated based on five representative indicators using a principal component analysis (PCA) method. It distinguishes and evaluates the fruit spots sensory differences between pear cultivars. The proposed method can effectively reduce the errors of manual observation, and improve the standardization of fruit spot phenotyping. It can also extend to other types of research of phenotypes with spots or textural characteristics. The data collected can help to establish standard databases of pear fruit phenotypes and conduct relevant omics studies.