1. Detection and Classification Defects on Exported Banana Leaves by Computer Vision
- Author
-
Nguyen Dao Xuan Hai, Duong Tan Dat, and Nguyen Truong Thinh
- Subjects
Visual processing ,Identification (information) ,Computer science ,business.industry ,Boundary (topology) ,Image processing ,Computer vision ,Artificial intelligence ,business - Abstract
Some effective techniques for identifying defects and estimating defect areas are the main requirements for computer vision and visual processing. This study provides an image processing algorithm to identify and calculate areas of defects on banana leaves. The algorithm consists of the main steps of processing images, segmenting images, labeling, size filtering, determining the boundaries for candidate areas such as chalks, spider webs, pus banana, soils, torn leave. Extracting colour characteristics to identify defects and estimate ultimately areas. Defect of banana has a lot of leak so we use number of defects to classify in good or reject leaf. Extracting boundary features and estimating boundary lengths to determine torn leaves. Experiments were conducted on 200 leaves to be identified. The accuracy of the proposed method is 89.8% for the method of color identification of disability defects and 94.7% for the method of identifying torn leaves.
- Published
- 2019
- Full Text
- View/download PDF