Year: 2018
Type: dissertation
Abstract: The ability to place, mark and evaluate a product in production lines is critical for the laser marking of horticultural plants. This study was divided into two objectives. Firstly, the investigation aimed to find the optimal position for marking horticultural products. The study developed and examined algorithms for determining the optimal marking position of poinsettia leaves and banana hands. Two methods of segmentation were performed: distance transform-watershed segmentation and main leaf skeleton segmentation. Secondly, laser marking parameters were evaluated in order to optimize the marking process without destroying the horticultural products (i.e. poinsettia leaves and bananas) and to ensure readable text, patterns or codes. Methods of image processing were employed to develop a laser positioning system and to read and assess the quality of the pattern or codes. According to the results, distance transform-watershed segmentation is effective in segmenting occluded plant leaves which show significant differences in depth. Using this method with Kinect V1 leads to a low identification rate. However, the precision of depth measurement can be enhanced by implementing Kinect V2, which yields a 92.5% classification rate, whereas the main leaf skeleton method yields a lower classification rate (80%). Moreover, the proposed algorithms are successful in placing a 2D code on banana hands effectively. Positioning of 2D codes on the ventral banana side is a suggested application. It is possible to use laser marking on poinsettia leaves and bananas without damaging the product. Laser marking of 0.5 W [0.55 J per character] proved to be the lowest energy that can produce a high contrast, and also produces no apparent heat affected zone that is visually detectable on the poinsettia leaves. In terms of laser marking on bananas, readability results are achieved by applying low power and a longer marking time, whereby an 80-100% readability of data matrix codes after storage was obtained.
DOI: 10.15488/4309
Author: Nasution, Indera Sakti
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
SDG: Zero hunger
Open Access status: green
OpenAlex ID: W2919845994