Open AccessResearch and Implementation of an OpenMV-Based Target Edge Detection and Tracking System
Author(s) -
Yige Guo,
Jun Yin,
Yu Wang,
Shikai Wu,
Le Luo,
Qiang Fu
Publication year - 2022
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2173/1/012083
Subject(s) - computer science , computer vision , python (programming language) , artificial intelligence , microcontroller , image processing , edge detection , kernel (algebra) , canny edge detector , tracking system , computer hardware , image (mathematics) , filter (signal processing) , operating system , mathematics , combinatorics
The improvement of ARM microcontroller unit (MCU) processing capability has greatly broadened the application fields of machine vision, and the huge benefits brought by machine vision can be seen in scenarios such as medicare and health, industrial production, military and national defense. Edge detection is a common processing method in image processing, which can provide basic guarantee for subsequent target tracking and other controls. In this paper, OpenMV is used as a hardware platform to extract the edges of a target in combination with specific algorithms, and then the common PID algorithm in the field of automatic control is written in the OpenMV integrated development environment (IDE) using Python language to achieve low-latency tracking of dynamic target objects. In addition, this thesis uses the color threshold tool in the IDE to capture the color threshold of defective product as an eigenvalue, and then calls the Canny operator in the function library and functions such as kernel filtering to process the image. Through actual testing and comparison, it can be found that the image edge is retained more completely after the kernel filtering process. After the general area of the defective products is calibrated on the object to be detected, the defective products can be screened.