Digital parameterization of apple fruit size, shape and surface spottiness

Abstract

To reach fruit market standards, quality evaluation has to be performed. Computer assisted fruit image analysis represents a technique, which offers a variety of automatic and semi-automatic procedures that can be used in combination with classic evaluation methods. To achieve this goal, a digital parameterization method for single apple fruit (Malus domestica) size, shape and surface spottiness has been recently developed. The appropriate mathematical procedures, defining the criteria for the fruit quality parameterization, are also defined and tested. The concept of the method, as well as the initial testing results, is presented in this paper. Basically, the technique combines analysis of apple fruit 256 gray-scale level images and parameterization algorithm of fruit quality. The former is based on digital pattern recognition method (DPR), and the latter employs linear fitting and numerical integration of DPR output data. This way, accurate parameterization of the fruit size, shape and surface spottiness, as well as the reliable fruit sorting according to the product quality, is enabled. Key words: Apple fruit, computer vision, quality criteria, mathematical procedure, digital pattern recognition.