研究目的
To develop an algorithm for detecting infestation in canary beans using UV-induced visible fluorescence and image processing to distinguish healthy from infested beans, reducing subjectivity in analysis.
研究成果
The algorithm achieved high specificity (99.78%) and sensitivity (90.70%), reducing subjectivity in bean infestation analysis. Future work should involve more samples, better cameras, and exploration of different UV wavelengths to improve accuracy.
研究不足
The use of Raspberry Pi limits computational load, preventing the development of an artificial intelligence system; accuracy could be improved with more samples and specialized cameras; different UV wavelengths might enhance identification.
1：Experimental Design and Method Selection:
The method involves acquiring images of beans under UV radiation in a hermetic enclosure, using image segmentation and histogram analysis algorithms to detect infestation.
2：Sample Selection and Data Sources:
1320 grains of canary beans with varying color and infestation presence were divided into 30 samples, evaluated by experts and the algorithm.
3：List of Experimental Equipment and Materials:
Hermetic enclosure, grid (15 x 25 cm), LifeCam Studio web cameras (2 units), Raspberry Pi computer, 6-W ultraviolet illumination (wavelength approximately 400 nm), vibrator, servomotor.
4：Experimental Procedures and Operational Workflow:
Beans are dosed onto a grid using an electromechanical system, images are captured from top and bottom under UV light, processed to extract R and G components, apply thresholding, labeling, filtering, and histogram analysis to identify infestation.
5：Data Analysis Methods:
Sensitivity and specificity calculations based on expert comparisons, using equations for true positives, false negatives, true negatives, and false positives.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
