Tomato Quality Identifier Applying Color-Based Segmentation Using MATLAB with K-Means Clustering and Pixel Area Subtraction
| Autor: | Engr. Leonardo A. Samaniego, Engr. Stanley Glenn E. Brucal, Patrick Vince L. Rodriguez, Johannah Kate A. Tolentino, Engr. Einstein D. Yong, Jolina G. Paz, Shaun Wesley I. Lemen |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
0106 biological sciences
Pixel business.industry Computer science Subtraction k-means clustering Sorting Image processing Pattern recognition 04 agricultural and veterinary sciences Image segmentation 01 natural sciences 040501 horticulture Segmentation Artificial intelligence 0405 other agricultural sciences business Cluster analysis 010606 plant biology & botany |
| Zdroj: | 2019 IEEE 11th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management ( HNICEM ). |
| DOI: | 10.1109/hnicem48295.2019.9072748 |
| Popis: | Sorting tomatoes before they are stored was done by several previous studies as damaged tomatoes tend to cause an increase in ripeness rate of other adjacent tomatoes which could lead to depletion of natural resources. However, most of these studies utilized only a single camera in inspecting the quality of tomato in which the damage is assumed to be facing the camera. Improvements will be done in order to detect not only one side but also the lateral side of the tomato. Along with this problem, there were also deficiencies in accuracy relating to damage detection and sorting. In this paper, the researchers have developed an accurate, wide coverage of detecting tomato surface, fast execution time, and a user-friendly tomato quality segregator. The study focuses on identifying the quality of tomato by getting the area of damage within a tomato. Three levels are introduced namely: healthy, slightly damaged, and heavily damaged tomatoes. MATLAB is used for the image processing. The method involves Color Based Segmentation which uses K-means clustering and Pixel Area Subtraction. Through series of testing, the researchers were able to design a system that has an accuracy for damage detection of 90.00% and an accuracy of 83.33% for segregation. The design project has a coverage area of detection of 95.24% for a tomato. It executes at an average time of 14.20 seconds from input to output process. The user-friendly rating of the system is 4.41. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|