Diseño de un prototipo de un sistema autónomo para determinar el precio de frutas basado en visión artificial.

Abstract

In the present work, a prototype of an autonomous system to determine the price of fruits ased on Artificial Vision was designed and built. It was divided into four modules to ensure its proper functioning. The first module is responsible for the weighing of fruit, and subsequent sending of the values to the computer. For this, a digital scale was built using of a load cell and a weight transmitter. An Arduino device was implemented for data transmission. The second module takes care of the acquisition of images in the best conditions by controlling the environment of the image capture. For this, a support was added to the digital scale where a circular lamp and a web cam in the center of it are located. In the third module, fruit recognition is carried out based on their color and characteristics in their morphology such as area, major axis and minor axis. In the fourth module, a graphical interface that unifies the previous stages was developed. The price to pay for the fruit is determined, and the user is provided with the purchase details autonomously, that is, without the need for human presence for this purpose. The prototype is able to recognize six different types of fruit, including green and red apples, green and yellow pears and yellow and red bananas. After a series of tests, it was determined that the system has an accuracy of 98% regarding the recognition of fruits.