基于改进差分进化算法的鳞翅目昆虫图像识别方法

doi:10.11707/j.1001-7488.20200308

Abstract

Abstract:

Objective: In this study, based on computer image processing technology to extract insect image features, a new feature selection technique was proposed to screen features related insect identification. The technique is developed for accurately and quickly identifying the species of Lepidopteran insects. Method: In order to improve the universality of the model, the open-source Leeds butterfly dataset and the dataset of the forest Lepidopteran insects were used in this study. The improved texture feature extraction algorithm (DRLBP) was used to extract the texture features of Lepidopteran insect images to obtain high recognition accuracy. The hamming distance was used to calculate the distance between particles and measure the population diversity. A method of automatically adjusting the diversity in the evolutionary process was proposed. A binary adaptive differential evolution (BADE) algorithm was presented. The BADE algorithm were used to select the appropriate subset of texture features with small dimensions, and probabilistic collaborative representation based classifier (PROCRC) was used to classify images. Result: PROCRC classifier showed good classification result on all datasets. Its average recognition rate was 81.73% and 88.18% respectively. The classification accuracy of insects was significantly improved by feature selection. Its highest rate of improvement was 13.49%. The performance of BADE algorithm were better than other feature selection algorithms. The dimension and classification time of texture dataset selected by the BADE algorithm was significantly reduced. The dimensionality reduction rate was close to 50%, and the time reduction rate was up to 50%. Conclusion: The BADE algorithm proposed in this study can effectively carry out feature selection, which improves the recognition accuracy and saves the recognition time of the model. It is indicated that the method of feature selection for Lepidopteran insect images using swarm intelligence optimization algorithm is feasible. In summary, the method of recognition of Lepidopteran Insects combined with DRLBP and BADE algorithm is proposed in this study, which has important application prospects in rapid and accurate identification of agricultural and forestry insects.

Key words: texture feature, binary adaptive differential evolution, feature selection, classifier

CLC Number:

S763.42

Dakun Lin,Shiguo Huang,Feiping Zhang,Guanghong Liang,Songqing Wu,xia Hu,Rong Wang. Method of Image Recognition for Lepidopteran Insects Based on Improved Differential Evolution Algorithm[J]. Scientia Silvae Sinicae, 2020, 56(3): 73-81.

Figures/Tables 9

Fig.1

Fig.2

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Fig.3

References 0

	蔡小娜, 韩旭, 沈佐锐, 等. 基于蛾翅翅脉特征的夜蛾昆虫数字化分类研究(鳞翅目:夜蛾科). 环境昆虫学报, 2016. 38 (2): 348- 353.
	Cai X N , Han X , Shen Z R , et al. Digital classification of noctuid moths (Lepidoptera:Noctuidae) base on wings vein characteristics. Journal of Environmental Entomology, 2016. 38 (2): 348- 353.
	黄世国. 2008.基于图像的昆虫识别关键技术研究.西安:西北大学博士学位论文.
	Huang S G. 2008. Research on the key techniques of image-based insects recognition. Xi'an: PhD thesis of Northwest University.[in Chinese]
	计智伟, 胡珉, 尹建新. 特征选择算法综述. 电子设计工程, 2011. 19 (9): 46- 51. doi: 10.3969/j.issn.1674-6236.2011.09.014
	Ji Z W , Hu M , Yin J X . A survey of feature selection algorithm. Electronic Design Engineering, 2011. 19 (9): 46- 51. doi: 10.3969/j.issn.1674-6236.2011.09.014
	李凡. 2015.基于数字图像的蝴蝶种类自动识别研究.北京:北京林业大学博士学位论文.
	Li F. 2015. The research on automatic identification of butterfly species based on the digital image. Beijing: PhD thesis of Beijing Forestry University.[in Chinese]
	李小林, 周蓬勃, 周明全, 等. 基于可区分二进制局部模式特征的蛾类昆虫识别. 计算机应用与软件, 2016. 33 (3): 172- 175. doi: 10.3969/j.issn.1000-386x.2016.03.039
	Li X L , Zhou P B , Zhou M Q , et al. Recogning insects of moth species based on discriminative local binary pattern features. Computer Applications and Software, 2016. 33 (3): 172- 175. doi: 10.3969/j.issn.1000-386x.2016.03.039
	薛薇. 统计分析与SPSS的应用. 北京: 中国人民大学出版社. 2011.
	Xue W . Statistical Analysis and SPSS Application. Beijing: China Renmin University Press. 2011.
	赵汗青, 沈佐锐, 于新文. 数学形态特征应用于昆虫自动鉴别的研究. 中国农业大学学报, 2002. 7 (3): 38- 42. doi: 10.3321/j.issn:1007-4333.2002.03.007
	Zhao H Q , Shen Z R , Yu X W . On computer-aided insect identification through math-morphology features. Journal of China Agricultural University, 2002. 7 (3): 38- 42. doi: 10.3321/j.issn:1007-4333.2002.03.007
	竺乐庆, 张真. 基于稀疏编码和SCGBPNN的鳞翅目昆虫图像识别. 昆虫学报, 2013. 56 (11): 1335- 1341. doi: 10.3724/SP.J.1105.2011.10014
	Zhu L Q , Zhang Z . Using sparse coding and SCGBPNN to recognize images of lepidopteran insects. Acta Entomologica Sinica, 2013. 56 (11): 1335- 1341. doi: 10.3724/SP.J.1105.2011.10014
	竺乐庆, 张大兴, 张真. 基于颜色名和OpponentSIFT特征的鳞翅目昆虫图像识别. 昆虫学报, 2015a. 58 (12): 1331- 1337.
	Zhu L Q , Zhang D X , Zhang Z . Recognition of lepidopteran species based on color name and opponent SIFT features. Acta Entomologica Sinica, 2015a. 58 (12): 1331- 1337.
	竺乐庆, 张大兴, 张真. 基于韦伯局部描述子和颜色直方图的鳞翅目昆虫翅图像特征描述与种类识别. 昆虫学报, 2015b. 58 (4): 419- 426.
	Zhu L Q , Zhang D X , Zhang Z . Feature description of lepidopteran insect wing images based on WLD and HoC and its application in species recognition. Acta Entomologica Sinica, 2015b. 58 (4): 419- 426.
	周爱明, 马鹏鹏, 席天宇, 等. 基于深度学习的蝴蝶科级标本图像自动识别. 昆虫学报, 2017. 60 (11): 1339- 1348.
	Zhou A M , Ma P P , Xi T Y , et al. Automatic identification of butterfly specimen images at the family level based on deep learning method. Acta Entomologica Sinica, 2017. 60 (11): 1339- 1348.
	周志华, 王珏. 机器学习及其应用2009. 北京: 清华大学出版社. 2009.
	Zhou Z H , Wang J . Machine learning and its applications. Beijing: Tsinghua University Press. 2009.
	Ashaghathra S M A , Weckler P , Solie J , et al. Identifying Pecan Weevils through image processing techniques based on template matching. American Society of Agricultural and Biological Engineering, Annual meeting, Mineapolis, Minnesota, June, 2007. 17- 20.
	Cai S , Zhang L , Zuo W , et al. A probabilistic collaborative representation based approach for pattern classification. Computer Vision and Pattern Recognition, IEEE, 2016. 2950- 2959.
	Krizhevshy A , Sutskever I , Hinton G . Imagenet classification with deep convolutional neural networks. Advances in Neural Information Processing Systems, 2012. 25, 1097- 1105.
	Larios N , Soran B , Shapiro L G , et al. Haar random forest features and SVM spatial matching kernel for stonefly species identification. International Conference on Pattern Recognition. IEEE Computer Society, 2010. 2624- 2627.
	Mehta R , Egiazarian K . Dominant rotated local binary patterns (DRLBP) for texture classification. Pattern Recognition Letters, 2016. 71, 16- 22. doi: 10.1016/j.patrec.2015.11.019
	Pampara G , Engelbrecht A P , Franken N . Binary differential evolution. 2006IEEE Congress on Evolutionary Computation, 2006. 1873- 1879.
	Storn R , Price K . Differential evolution-a simple and efficient heuristic for global optimization over continuous spaces. Journal of Global Optimization, 1997. 11 (4): 341- 359. doi: 10.1023/A:1008202821328
	Tofilski A . DrawWing, a program for numerical description of insect wings. Journal of Insect Science, 2004. 4 (17): 1- 5. doi: 10.1673/031.004.1701
	Wang J, Markert K, Everingham M. 2009. Learning models for object recognition from natural language descriptions. British Machine Vision Conference, BMVC 2009, London, UK, September 7-10, 2009.
	Yang H P , Ma C S , Wen H , et al. A tool for developing an automatic insect identification system based on wing outlines. Scientific Reports, 2015. 5, 12786. doi: 10.1038/srep12786
	Zhang L , Yang M , Feng X . Sparse representation or collaborative representation:which helps face recognition? International Conference on Computer Vision. IEEE Computer Society, 2011. 471- 478.

数据集 Datasets	分类器 Classifiers	不同运行次数的识别率 Recognition rates of different running times(%)					均值 Average	显著性水平 Significance level
数据集 Datasets	分类器 Classifiers	1	2	3	4	5	均值 Average	显著性水平 Significance level
利兹Leeds	PROCRC	82.08	81.13	81.61	82.21	81.61	81.73	＜0.001
利兹Leeds	KNN	38.59	40.38	39.68	39.30	40.13	39.61	＜0.001
本课题组 Taken in this study	PROCRC	88.25	88.30	88.02	88.70	87.62	88.18	＜0.001
本课题组 Taken in this study	KNN	84.05	83.99	83.76	84.56	83.02	83.87	＜0.001

数据集 Datasets	种类 Species	样本数量 Number of samples	PROCRC识别率 Recognition rate of PROCRC(%)	KNN识别率 Recognition rate of KNN(%)	显著性水平 Significance level
利兹Leeds	黑脉金斑蝶Danaus plexippus	82	95.07	37.65	＜0.001
	黄条袖蝶Heliconius charitonius	93	97.89	80.53	0.01
	红带袖蝶Heliconius erato	61	64.23	0.00	＜0.001
	鹿眼蛱蝶Junonia coenia	90	70.00	45.56	0.01
	红灰蝶Lycaena phlaeas	88	79.48	47.52	＜0.001
	丧服蛱蝶Nymphalis antiopa	100	82.00	47.00	＜0.001
	美洲大芷凤蝶Papilio cresphontes	89	77.39	41.44	＜0.001
	白粉蝶Pieris rapae	55	98.18	72.73	＜0.001
	优红蛱蝶Vanessa atalanta	90	64.44	10.00	＜0.001
	小红蛱蝶Vanessa cardui	84	91.69	13.09	＜0.001

本课题组 Taken in this study	檗黄粉蝶Eurema blanda	7	50.00	20.00	0.35
	茶蚕蛾Andraca bipunctata	238	95.39	85.28	＜0.001
	东方粉蝶Pieris canidia	12	23.33	0.00	0.08
	斐豹蛱蝶Argynnis hyperbius	5	20.00	0.00	0.37
	黄星尺蛾Arichanna melanaria fraternal	167	91.59	91.60	1.00
	灰白蚕蛾Ocinara varians	200	99.50	100.00	0.37
	蕾鹿蛾Amata germana	16	88.33	51.67	0.09
	柳杉毛虫Dendrolimus houi	372	96.23	91.12	0.01
	绿尾大蚕蛾Actias selene ningpoana	9	100.00	80.00	0.18
	马尾松毛虫Dendrolimus punctatus	359	82.46	80.50	0.62
	榕透翅毒蛾Perina nuda	349	83.97	80.51	0.39
	长喙天蛾Macroglossum corythus luteata	27	14.67	3.33	0.05

数据集 Datasets	分类器 Classifier	不同特征选择方法的识别率 Recognition rate of different feature selection approaches(%)
数据集 Datasets	分类器 Classifier	基于频繁出现模式 Based on the most frequently occurring patterns	二进制差分进化 Binary differential evolution	二进制自适应差分进化 Binary adaptive differential evolution
利兹Leeds	PROCRC	65.92	85.02	87.54
利兹Leeds	KNN	52.25	49.54	53.10
本课题组 Taken in this study	PROCRC	84.55	88.86	89.84
本课题组 Taken in this study	KNN	84.11	86.16	86.96

数据集 Datasets	种类 Species	样本数量 Number of samples	识别率 Recognition rate (%)		显著性水平 Significance level
数据集 Datasets	种类 Species	样本数量 Number of samples	PROCRC	KNN	显著性水平 Significance level
利兹Leeds	黑脉金斑蝶Danaus plexippus	82	97.5	67.06	＜0.001
	黄条袖蝶Heliconius charitonius	93	94.62	87.13	0.11
	红带袖蝶Heliconius erato	61	72.18	9.74	＜0.001
	鹿眼蛱蝶Junonia coenia	90	80	56.67	＜0.001
	红灰蝶Lycaena phlaeas	88	84.18	64.9	0.02
	丧服蛱蝶Nymphalis antiopa	100	85	51	0.02
	美洲大芷凤蝶Papilio cresphontes	89	85.42	45.95	0.01
	白粉蝶Pieris rapae	55	98.18	70.91	0.01
	优红蛱蝶Vanessa atalanta	90	58.89	22.22	＜0.001
	小红蛱蝶Vanessa cardui	84	87.87	22.57	＜0.001

本课题组 Taken in this study	檗黄粉蝶Eurema blanda	7	50	40	0.40
	茶蚕蛾Andraca bipunctata	238	92.87	84.45	0.09
	东方粉蝶Pieris canidia	12	10	0	0.18
	斐豹蛱蝶Argynnis hyperbius	5	0	0	—
	黄星尺蛾Arichanna melanaria fraternal	167	93.44	92.21	0.87
	灰白蚕蛾Ocinara varians	200	99	99.5	1.00
	蕾鹿蛾Amata germana	16	86.67	68.33	0.13
	柳杉毛虫Dendrolimus houi	372	96.5	92.19	＜0.001
	绿尾大蚕蛾Actias selene ningpoana	9	100	100	0.37
	马尾松毛虫Dendrolimus punctatus	359	82.74	80.23	0.33
	榕透翅毒蛾Perina nuda	349	83.94	82.24	0.94
	长喙天蛾Macroglossum corythus luteata	27	3.33	14.67	0.92

数据集 Datasets	分类器 Classifier	不同特征选择的F1值F1-score of different feature selection methods(%)
数据集 Datasets	分类器 Classifier	基于频繁出现模式 Based on the most frequently occurring patterns	二进制差分进化 Binary differential evolution	二进制自适应差分进化 Binary adaptive differential evolution
利兹Leeds	PROCRC	61.18	85.33	87.29
利兹Leeds	KNN	50.86	48.69	53.92
本课题组 Taken in this study	PROCRC	83.61	88.66	89.03
本课题组 Taken in this study	KNN	83.59	85.53	86.67

Method of Image Recognition for Lepidopteran Insects Based on Improved Differential Evolution Algorithm

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 9

References 0

Related Articles 9

Recommended Articles

Metrics

Comments

数据集 Dataset	样本数 Sample number	原始维数 Original dimension	经特征选择(PROCRC) Feature selection(PROCRC)	经特征选择(KNN) Feature selection(KNN)
利兹 Leeds	832	768	399.6	344.2
本课题组 Taken in this study	1761	768	425.4	378.3

[1]	Zhulin Chen,Xuefeng Wang. Prediction of Total Iron Content in Dalbergia odorifera Leaves Based on Vegetation Index and Multispectral Texture Parameters [J]. Scientia Silvae Sinicae, 2020, 56(2): 89-98.
[2]	Song Yining, Liu Wenping, Luo Youqing, Zong Shixiang. Monitoring of Dead Trees in Forest Images Based on Linear Spectral Clustering [J]. Scientia Silvae Sinicae, 2019, 55(4): 187-195.
[3]	Liu Luxia, Pang Yong, Ren Haibao, Li Zengyuan. Predict Tree Species Diversity from GF-2 Satellite Data in a Subtropical Forest of China [J]. Scientia Silvae Sinicae, 2019, 55(2): 61-74.
[4]	Han Zongtao, Jiang Hong, Wang Wei, Li Zengyuan, Chen Erxue, Yan Min, Tian Xin. Forest Above-Ground Biomass Estimation Using KNN-FIFS Method Based on Multi-Source Remote Sensing Data [J]. Scientia Silvae Sinicae, 2018, 54(9): 70-79.
[5]	Zhang Ying, Zhang Xiaoli, Li Hongzhi, Li Liangcai. A Comparison of Landcover Classification Based on the Improved Transformed Divergence Analysis [J]. Scientia Silvae Sinicae, 2018, 54(8): 88-98.
[6]	Guo Hui, Wang Xiao, Liu Chuanze, Zhou Yucheng. Research on Defect Extraction of Particleboard Surface Images Based on Gray Level Co-Occurrence Matrix and Hierarchical Clustering [J]. Scientia Silvae Sinicae, 2018, 54(11): 111-120.
[7]	Ren Chong, Ju Hongbo, Zhang Huaiqing, Huang Jianwen, Zheng Yingxuan. Multi-Source Data for Forest Land Type Precise Classification [J]. Scientia Silvae Sinicae, 2016, 52(6): 54-65.
[8]	Li Chungan;Shao Guofan. Object-Oriented Classification of Forest Cover Using SPOT5 Imagery [J]. Scientia Silvae Sinicae, 2010, 46(8): 130-139.
[9]	Yu Haipeng;Liu Yixing;Zhang Bin;Li Yongfeng. Application of Spatial Gray Level Cooccurrence Matrix in Wood Surface Texture Quantitative Analysis [J]. Scientia Silvae Sinicae, 2004, 40(6): 121-129.