Automatic Grader for Flatfishes Using Machine Vision
- PDF / 2,165,546 Bytes
- 9 Pages / 594.77 x 793.026 pts Page_size
- 81 Downloads / 233 Views
ISSN:1598-6446 eISSN:2005-4092 http://www.springer.com/12555
Automatic Grader for Flatfishes Using Machine Vision Hee-Jee Sung, Myeong-Kwan Park*, and Jae Weon Choi Abstract: Labor shortage and an aging population are driving automation in the aqua and agricultural industries. In a flatfish farm, it is important to sort flatfishes according to their sizes for effective and stable growth. This sorting is often done by human eye-estimation, thereby making it difficult to carry out total inspection of the fishes on the farm. Most fish graders in the industries sort round fishes (fishes with high length to height ratios) but they are inadequate to sort flatfishes. Thus, in this study, an automatic grader for flatfish using machine vision is developed. The grader has three main parts: conveyor belt, machine vision, and sorter. The conveyor belt transfers the fishes to the measurement and sorter parts. When the fish is detected and its length is calculated by image processing, the position of the sorter is controlled by length classification. A low-cost commercial webcam is used, and the sorter of the grader has a simple structure that consists of a single actuator. After several experiments, it was verified that length measurement using machine vision is accurate to within 10 mm, and the grader can sort 30 fishes in a minute. The developed grader minimizes the out-of-water exposure of the fishes compared to the conventional eye-measurement. Hence, it is effective to maintain the quality and freshness of the fishes. Keywords: Aquaculture, fish grader, flatfish, image processing, machine vision.
1.
INTRODUCTION
The social issue of labor shortage in the farming and fishery industries is becoming critical [1]. Thus, many studies for automating agriculture and aquaculture have been performed worldwide. For example, a weight sort is an important process in farming, and in aquacultural and agricultural industries, there have been many studies on the weight sort of products based on weight measuring sensors, such as a loadcell and a pressure sensor. A weight sorting system utilizing load cell was developed to grade dried-persimmon [2]. Additionally, a weightgrading machine for oranges was developed [3]. An automated weight sorting system for fruits has also been designed to utilize a pressure sensor [4]. Due to electric and mechanical noises from moving parts, such as the conveyor belt, some studies have focused on noise reduction techniques. To enhance the accuracy of a loadcell-based weighing machine used in a conveyor belt, a filter slope control was developed, which uses a selective switching of multiple low pass filter stages to form optimum pulse width ratio [5]. A modified auto-regressive model was also proposed for a stable prediction of loadcell signal of a high-speed weighing machine [6]. However, a moving object, such as a live fish, is hard to
sort by conventional weight measuring sensors, such as a loadcell and a pressure sensor. Thus, many manufacturers of grading machines for live fish have preferred sensorl
Data Loading...
