The Design And Implementation Of Vegetables Recognition Module Based On Computer Vision

In recent years, the process of agricultural commercialization was promoted constantly in our country. However, the huge production of vegetable poses challenges to the products market supervision and supply chain management. At present, most of the sorting of agricultural products managements are still rely on labor, which not only cost a lot of labor, and the efficiency is low, causing serious influence upon the development speed of commercialization of fruits and vegetables. In this paper, the design of vegetables recognition module based on computer vision technology was proposed, which will free the labor personnel from onerous agricultural management, improve labor efficiency.Computer vision technology is a kind of comprehensive technology which uses an image capture unit and processing unit, instead of human vision, carrying out some operations such as image segmentation and classification, tracking, analysis of the statistical law of image features. For the issue that traceability scale PLU code was input incorrectly, leads to a problem that the traceability function was not carried out. In this paper, the design and implementation of automatic identification module for vegetables was established, provides an automation solutions for traceability scale system.The main idea of this paper is established a vegetable recognition module including hardware setup and supporting software process, which carry out vegetable image acquisition, segmentation, feature extraction and classification and identification functions. The module is meant to replace the traceability scale PLU manual input.The main work is as follows:1. Investigation of the researches that scholars at home and abroad study in the field of agricultural product identification, to analyze the feasibility of the plan.2. According to the specific of the scale of the functional requirements, complete the identification module of hardware design and build, software process design, etc.3. Development of corresponding software functions, including image capture, serial port communication, image preprocessing and image segmentation, feature extraction and recognition, and completing implementation of vegetables identification process.4. In the end, a test that testing 24 kinds of seasonal vegetables, validating indicators of the module, was posed to analyze its feasibility and accuracy.This paper introduces the design and implantation of the "vegetable recognition module" project, using a classic algorithm which contains image features and KNN classifier. The module is running in CortexA9 architecture i.MX6Q development board. The powerful hardware performance guarantee the speed of image processing. The whole system running time is about 2 seconds, meeting the demand of application scenarios. In the end, this module implements the function of automatic identification of vegetables efficiently and quickly, all parts work stability, which can realize image acquisition, processing, integration operation such as identification, delivering results show that, establishing an effective and feasible platform for the recognition of vegetables.