Current Status and Future Perspective of Spectroscopy and Imaging Technique Applications in Bruise Detection of Fruits and Vegetables: A Review

Tong-tong ZHOU; Xiao-lin SUN; Zhi-zhong SUN; He-huan PENG; Tong SUN; Dong HU

doi:10.3964/j.issn.1000-0593(2022)09-2657-09

Journals >Spectroscopy and Spectral Analysis >Volume 42 >Issue 9 >Page 2657 > Article

Spectroscopy and Spectral Analysis
Vol. 42, Issue 9, 2657 (2022)

Current Status and Future Perspective of Spectroscopy and Imaging Technique Applications in Bruise Detection of Fruits and Vegetables: A Review

Tong-tong ZHOU^1、*, Xiao-lin SUN^1、1;, Zhi-zhong SUN^2、2;, He-huan PENG^1、1;, Tong SUN^1、1;, and Dong HU^{1、1; *;}

Author Affiliations

¹1. College of Optical, Mechanical and Electrical Engineering, Zhejiang Agricultural and Forestry University, Hangzhou 311300, China

²2. College of Mathematics and Computer Science, Zhejiang Agricultural and Forestry University, Hangzhou 311300, China

show less

DOI: 10.3964/j.issn.1000-0593(2022)09-2657-09 Cite this Article

Tong-tong ZHOU, Xiao-lin SUN, Zhi-zhong SUN, He-huan PENG, Tong SUN, Dong HU. Current Status and Future Perspective of Spectroscopy and Imaging Technique Applications in Bruise Detection of Fruits and Vegetables: A Review[J]. Spectroscopy and Spectral Analysis, 2022, 42(9): 2657 Copy Citation Text

show less

Fig. 1. Schematic diagram of fluorescence imaging system^[32]

Download full size | View in the Article

Fig. 2. (a) Hyper-spectral images of pear before and after normalization at 1 200 nm; (b) Average spectra of pear at 950~1 650 nm^[43]

Download full size | View in the Article

Fig. 3. Schematic diagram of spatial-frequency domain imaging system

Download full size | View in the Article

检测技术		优点	缺点
光谱技术	近红外光谱^[4]	无损环保、简单快速、易于操作	建模复杂、模型通用性弱、不具备空间信息
	拉曼光谱^[5]	准确性高、无损快速高效、具有更窄、更清晰的分子峰特征	样品准备复杂、无法实时获取信息
	荧光光谱^[6]	高灵敏度、快速、无损、装置成本低	存在错峰重叠、归属不明的问题
成像技术	机器视觉^[7]	效率高、灵活性高、工作时间长	无法获取物体内部信息
	高光谱成像^[8]	图谱合一, 具备光谱和空间信息	数据量大、特征波段的选择和准确性不稳定
	空间频域成像^[9]	深度辨析、信号增强	需要选择特征波段、实时性不高
	磁共振成像^[10]	快速直观, 能得到空间信息和不同切层图像信息	设备成本较高、成像速度较慢
	X射线成像^[10]	穿透能力强, 能反应内部特征	成本相对较高, 对安装及安全要求严格
	热成像^[11]	成像速度快、检测面积大	对比度弱、信噪比低、高度依赖环境条件

Table 1. Comparison of the characteristics of different spectroscopy and imaging techniques for bruising detection of fruits and vegetables

View in the Article

检测技术	检测对象		损伤类型			数据处理算法	正确率/%		文献
近红外光谱	苹果		擦伤冷害			PLS-DA ANN, SVM	94.0~96.0 98.3		[20] [22]
	花椰菜		腐烂			PLS	100		[25]
	鸭梨		黑心病黑心病			PLS PLS	100 98.3		[23] [24]
	梨枣		碰伤			PLS-LDA	96.7		[14]
	龙眼		碰伤			PLS-DA	100		[15]
	猕猴桃		碰伤			SPA-LSSVM	98.2		[13]
	橄榄		碰伤			PLS	100		[16]
	樱桃		碰伤			LSSVM	93.3		[18]
	番茄		碰伤			LSSVM	98.4		[17]
拉曼光谱	橄榄		冻伤、发酵			SIMCA, PLS-DA, K-NN	100, 97.0		[27]
荧光光谱		苹果土豆		擦伤擦伤	PCA PCA			- -	[31] [31]
荧光成像		苹果		碰伤	M-value			0.5 h: 86.7 1 h后: 100	[32]
高光谱成像		青椒		冷害	PLS-DA			84.0	[56]
		苹果		碰伤瘀伤擦伤碰伤	iPLS-DA WS SVM SVM			92.4, 94.0 99.5 97.5 97.3	[51] [52] [53] [54]
		桃子		擦伤	PCA, WS			96.6	[36]
				早期擦伤	PCA, WS			96.5(损伤果) 97.5(健康果)	[39]
				冷害	SVM, ANN			99.3(二分类) 96.1(三分类) 85.4(四分类)	[59]
		番茄		开裂	LDA, SVM			94.6, 96.4	[42]
		蓝莓		瘀伤	CARS-LSSVM			12 h后: 95.0	[48]
		蓝莓		瘀伤	SVM			94.0(训练集) 92.0(测试集)	[41]
		樱桃		冷害	BPNN			83.3(冷冻果) 94.6(健康果)	[49]
		柑橘		早期腐烂	PCA, WS			100(训练集) 98.6(测试集)	[37]
		梨		瘀伤	F-value			92.0	[44]
		马铃薯		黑心病	PLS-DA			94.0	[55]
		茄子		冷害	SVM			100	[50]
		青枣		冷害	LDA			98.3(二分类) 93.3(三分类)	[44]
		黄瓜		冷害	SVM			100(二分类) 90.5(三分类)	[45]
空间频域成像		苹果		碰伤碰伤碰伤	TPD SPT Otsu			70~100 - >85.8	[60] [61] [62]
磁共振成像		鳄梨梨苹果		瘀伤损伤体积水心病	- DL -			- - 100	[65] [66] [75]
X射线成像		苹果		内部褐变内部褐变	- -			- -	[69] [71]
X射线成像		石榴		内部损伤	-			-	[70]
热成像		梨蓝莓		碰伤损伤	A photothermal model LDA, SVM, RF, K-NN等			- 89.5	[72] [73]

Table 2. Application status of spectroscopy and imaging techniques for bruising detections of fruits and vegetables

Download Citation

Save the article for my favorites

Paper Information