摘要
高光谱成像技术包含图像信息和光谱信息。本文利用高光谱成像技术检测苹果摔伤,主要采用主成分分析、波段比算法和支持向量机分析所采集的高光谱图像数据。实验结果表明,波段比算法和主成分分析法分类识别正确率为93.3%,与支持向量机相比更适用于苹果摔伤的实时快速检测。
Hyper-spectral imaging technology includes image information and spectral information. This paper usedhyperspectral imaging technology to detect apple fall. In the process of experiment, principal component analysis,band ratio algorithm and support vector machine were used to analyze hyperspectral image data collected. The experi-mental results showed that the accuracy of band ratio algorithm and principal component analysis was 93. 3%, whichwas more suitable for real time and fast detection of apple fall than support vector machine.
引文
[1]赵娟,彭彦坤,赵松玮,等.基于高光谱技术检测苹果外观缺陷[J].食品安全质量检测学报,2012(6):681-684.
[2]索少增,刘翠玲,吴静珠,等.高光谱图像技术检测梨表面农药残留试验研究[J].北京工商大学学报(自然科学版),2011(6):73-77.
[3]彭谨.主成分分析入门[EB/OL].(2016-10-16)[2018-03-01].https://www.cnblogs.com/SCUJIN/p/5965946.html.
[4]张明月,贺金兰,田尉霞.主成分分析在高光谱图像降维中的应用[J].内蒙古科技与经济,2017(15):88-90.
[5]蔡健荣,王建黑,陈全胜,等.波段比算法结合高光谱图像技术检测柑橘果锈[J].农业工程学报,2009(1):127-131.
[6]Inoue Y,Pe?uelas J,Miyata A,et al.Normalized difference spectral indices for estimating photosynthetic efficiency and capacity at a canopy scale derived from hyperspectral and CO2flux measurements in rice[J].Remote Sensing of Environment,2008(1):156-172.
[7]Steve De Backer,Aleksandra Pi?urica,Bruno Huysmans,et al.Denoising of multicomponent images using wavelet least-squares estimators[J].Image and Vision Computing,2008(7):1038-1051.
[8]Thomas Koenig,Lester Melie-Garcia,Maria Stein,et al.Establishing correlations of scalp field maps with other experimental variables using covariance analysis and resampling methods[J].Clinical Neurophysiology,2008(6):1262-1270.
[9]百度百科.支持向量机[EB/OL].(2015-01-13)[2018-03-01].https://baike.baidu.com/item/%E6%94%AF%E6%8C%81%E5%90%91%E9%87%8F%E6%9C%BA/9683835.
[10]常用的核函数[EB/OL].(2016-07-03)[2018-03-01].https://blog.csdn.net/qq_27231343/article/details/51817866.
[11]赵杰文,刘剑华,陈全胜,等.利用高光谱图像技术检测水果轻微损伤[J].农业机械学报,2008(1):106-109.
[12]陈全胜,赵杰文,蔡健荣,等.利用高光谱图像技术评判茶叶的质量等级[J].光学学报,2008(4):669-674.