Methods for Identifying and Classifying Quality of Passion Fruit Based on Multi Feature Fusion
Passion fruit sales need to be classified and priced based on quality including appearance and taste.Recently,the quality assessment is a manual operation mainly relying on human eye judgment and lack of quantitative standards,which is easily influenced by visual fatigue and subjective experience.We propose a multi-feature fusion method based on computer vision to automatically assess the quality of passion fruit.Extract the color and texture apparent features of passion fruit separately for fusion and optimization first;then an SVM classifier is used to perform binary correlation training and prediction between the fused features and the fruit fructose content.The highest classification accuracy of the experimental results can reach 91.48%,indicating a significant correlation between the appearance characteristics of passion fruit and its taste quality and using machine vision to quality recognition of passion fruit and other fruits has high feasibility.The algorithm used is simple,fast and effective,and with great application value.
万方数据
维普
