研究目的
Investigating the application of hyperspectral imaging for detecting surface defects on jujube fruits, specifically focusing on the 'Kaohsiung 11' variety.
研究成果
Hyperspectral imaging combined with PCA and band ratio methods is effective for detecting surface defects on jujube fruits. The study identified optimal wavelengths for defect detection and proposed methods to differentiate rusty regions from other defects. Future research will focus on integrating PCA on selected wavelengths and developing a rotational hyperspectral scanning system for on-line detection.
研究不足
The glare due to specular reflection from the smooth and waxy surface of jujube may lead to errors in differentiating surfaces with defects from sound surfaces. Only the upper half of the jujube's surface can be scanned in one push broom scan, limiting the area analyzed in a single scan.
1：Experimental Design and Method Selection:
Developed a hyperspectral imaging system to acquire reflection spectra of jujube surfaces with defects. Used Principal Component Analysis (PCA) to reduce spectral dimensionality and identify significant wavebands for defect detection.
2：Sample Selection and Data Sources:
Collected jujube samples with normal surfaces and four common defective surface conditions (decay, rusty spots, fungal infection, and insect bites) from a commercial orchard in Pingtung, Taiwan.
3：List of Experimental Equipment and Materials:
Hyperspectral imaging system components included a Basler ace acA1920-155um monochrome camera, Imaging spectrograph (Imspector V10, Spectral Imaging Ltd.), 50 mm focal length lens, two halogen lamps, and a linear stage driven by stepper motor.
4：Experimental Procedures and Operational Workflow:
Acquired hyperspectral images of jujube samples, applied PCA to the image data, and used band ratio methods for defect detection. Calibrated images with relative reflectance for further analysis.
5：Data Analysis Methods:
Analyzed hyperspectral image data using GUI programs developed in Labview and Matlab environments. PCA was applied to reduce dimensionality and identify dominant wavelengths for defect detection.
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
