Pest collection method based on global contrast
-
摘要:
目的 目前利用测报灯，通过灯光诱捕昆虫，并由计算机完成昆虫图像的采集、计数和识别已逐步成为害虫测报的重要方法。为了减少昆虫在采样盘上重叠造成的计数和识别误差，基于害虫图像，根据昆虫密度研究采样盘中昆虫的收集方法，从而提高采集效率和精度。 方法 根据昆虫在采样盘上姿态特点，提出基于全局对比度的图像分割方法，结合阈值迭代分割获得昆虫区域，计算昆虫比例，并控制采样盘翻转完成对昆虫的收集。 结果 通过对5种害虫的实际图像进行的试验表明：与水平集、大津法(OTSU)、阈值迭代法和基于直方图对比度的显著性检测(HC)4种算法相比，本研究方法在准确率和召回率上均提高10%以上，取得了较好的结果；同时，在分割速度上比水平集快3倍，与阈值和HC算法基本持平。 结论 基于全局对比度的分割方法简单、高效，在害虫自动测报中具有较高的实际应用价值。图7表1参17 Abstract:Objective The real-time and accurate forecast of field pests has gradually become an important method of pest forecasting to trap insects by light, and to collect, count and identify insect images by computer. This study aims to explore the method of collecting insects in the sampling plate based on insect image and insect density, so as to improve the collection efficiency and accuracy, and reduce the counting and identification errors caused by the overlapping of insects on the sampling plate. Method According to the characteristics of insects’ posture on the sampling plate, an image segmentation method based on global contrast was proposed. Combined with threshold iterative segmentation, the insect area was obtained, the insect proportion was calculated, and the collection of insects was completed by controlling the flip of the sampling plate. Result Experiments on the actual images of 5 kinds of pest species showed that compared with the 4 algorithms, namely level set, OTSU, threshold iteration and saliency detection based on histogram contrast(HC), the accuracy and recall rate of this method were improved by more than 10%, and good results were obtained. At the same time, the segmentation speed was 3 times faster than that of the level set, which was basically the same as the threshold and HC algorithm. Conclusion Due to its simplicity and high efficiency, the segmentation method based on global contrast has high practical application value in automatic pest detection and reporting. [Ch, 7 fig. 1 tab. 17 ref.] -
Key words:
- forest protection /
- pest forecasting /
- pest collection /
- image segmentation /
- sampling disk /
- global contrast
-
图 4 OTSU算法与本研究算法对阴影分割的比较
Figure 4 Comparison of shadow segmentation between OTSU and research algorithm
图 5 HC算法与本研究算法对害虫细节分割的比较
Figure 5 Comparison of HC algorithm and research algorithm on pest detail segmentation
图 7 HC算法与本研究算法对深色害虫分割的比较
Figure 7 Comparison of HC algorithm and research algorithm on dark pest segmentation
表 1 准确率、召回率和分割速度对比
Table 1. Comparison of accuracy recall, and segmentation speed
分割算法 准确率 召回率 分割速度 白底样本 红底样本 白底样本 红底样本 白底样本 红底样本 水平集算法 0.72 0.74 0.62 0.64 1.68 4.35 OTSU算法 0.70 0.77 0.63 0.70 0.36 0.84 阈值迭代算法 0.83 0.83 0.81 0.69 0.56 0.87 HC算法 0.86 0.88 0.86 0.63 0.43 0.98 本研究算法 0.92 0.95 0.95 0.89 0.66 1.05 
下载: 导出CSV
-
[1] 张芳群. 基于Web技术的虫害预测系统的研究[D]. 杭州: 浙江理工大学, 2017. ZHANG Fangqun. Research on the Pest Forecasting System based on Web Technology[D]. Hangzhou: Zhejiang Sci-Tech University, 2017. [2] 陈梅香, 杨信廷, 石宝才, 等. 害虫自动识别与计数技术研究进展与展望[J]. 环境昆虫学报, 2015, 37(1): 176 − 183. CHEN Meixiang, YANG Xinting, SHI Baocai, et al. Research progress and prospect of technologies for automatic identifying and counting of pests [J]. J Environ Entomol, 2015, 37(1): 176 − 183. [3] 张书平, 余燕, 毕守东, 等. 灾变模型在马尾松毛虫幼虫发生量预报中的应用[J]. 浙江农林大学学报, 2020, 37(1): 93 − 99. doi: 10.11833/j.issn.2095-0756.2020.01.012 ZHANG Shuping, YU Yan, BI Shoudong, et al. A catastrophe model to forecast larvae occurrence of Dendrolimus punctatus [J]. J Zhejiang A&F Univ, 2020, 37(1): 93 − 99. doi: 10.11833/j.issn.2095-0756.2020.01.012 [4] 张鸣华, 丁鑫, 刘持标. 基于二值化图像识别的新型农业害虫诱杀监测系统研究[J]. 南昌航空大学学报(自然科学版), 2019, 33(1): 74 − 79. ZHANG Minghua, DING Xin, LIU Chibiao. Research on new agricultural pest trapping and killing monitoring system based on binarized image recognition [J]. J Nanchang Hangkong Univ Nat Sci, 2019, 33(1): 74 − 79. [5] 陈京, 刘德营, 谢堂胜, 等. 田间稻飞虱图像远程实时采集系统的研制[J]. 湖南农业大学学报(自然科学版), 2016, 42(6): 693 − 698. CHEN Jing, LIU Deying, XIE Tangsheng, et al. A remote real-time acquisition system for rice plant hopper images in the fields [J]. J Hunan Agric Univ Nat Sci, 2016, 42(6): 693 − 698. [6] SOLIS-SÁNCHEZ L O, GARCÍA-ESCALANTE J J, CASTAÑEDA-MIRANDA R, et al. Machine vision algorithm for whiteflies(Bemisia tabaci Genn)scouting under greenhouse environment [J]. J Appl Entomol, 2009, 133(7): 546 − 552. doi: 10.1111/j.1439-0418.2009.01400.x [7] NING Jifeng, ZHANG Lei, ZHANG David, et al. Interactive image segmentation by maximal similarity based region merging [J]. Pattern Recognition, 2009, 43(2): 445 − 456. [8] 吕金娜. 基于LAB空间和自适应聚类的害虫图像分割方法[J]. 河南科技学院学报(自然科学版), 2016, 44(1): 57 − 61. LÜ Jinna. Image segmentation based on LAB color space adaptive clustering algorithm for pest image segmentation [J]. J Henan Inst Sci Technol Nat Sci Ed, 2016, 44(1): 57 − 61. [9] 杨信廷, 刘蒙蒙, 许建平, 等. 自动监测装置用温室粉虱和蓟马成虫图像分割识别算法[J]. 农业工程学报, 2018, 34(1): 164 − 170. doi: 10.11975/j.issn.1002-6819.2018.01.22 YANG Xinting, LIU Mengmeng, XU Jianping, et al. Image segmentation and recognition algorithm of greenhouse whitefly and thrip adults for automatic monitoring device [J]. Trans Chin Soc Agric Eng, 2018, 34(1): 164 − 170. doi: 10.11975/j.issn.1002-6819.2018.01.22 [10] 陈树越, 吴正林, 朱军, 等. 基于凹点检测的粮仓粘连害虫图像分割算法[J]. 计算机工程, 2018, 44(6): 213 − 218. doi: 10.3969/j.issn.1000-3428.2018.06.037 CHEN Shuyue, WU Zhenglin, ZHU Jun, et al. Image segmentation algorithm for grain overlapping pests based on pitting detection [J]. Comput Eng, 2018, 44(6): 213 − 218. doi: 10.3969/j.issn.1000-3428.2018.06.037 [11] 李芝茹, 吴晓峰, 李全罡, 等. 自动追踪式可升降太阳能虫害监测装置的设计[J]. 森林工程, 2016, 32(6): 85 − 88. doi: 10.3969/j.issn.1001-005X.2016.06.017 LI Zhiru, WU Xiaofeng, LI Quangang, et al. Design of the solar liftable device for pest monitoring with automatic trac-king [J]. For Eng, 2016, 32(6): 85 − 88. doi: 10.3969/j.issn.1001-005X.2016.06.017 [12] 张红涛, 胡玉霞, 刘新宇, 等. 农田害虫实时检测装置的设计与实现[J]. 河南农业科学, 2007(12): 63 − 65. doi: 10.3969/j.issn.1004-3268.2007.12.018 ZHANG Hongtao, HU Yuxia, LIU Xinyu, et al. Design and implementation of real-time detection device for farmland pests [J]. J Henan Agric Sci, 2007(12): 63 − 65. doi: 10.3969/j.issn.1004-3268.2007.12.018 [13] 张昊辰, 申小艳, 张欣铃. 智能粮仓害虫监测系统的设计[J]. 通信电源技术, 2013, 30(4): 111 − 112, 126. doi: 10.3969/j.issn.1009-3664.2013.04.036 ZHANG Haochen, SHEN Xiaoyan, ZHANG Xinling. Design of smart granary pests monitoring system [J]. Telecom Power Technol, 2013, 30(4): 111 − 112, 126. doi: 10.3969/j.issn.1009-3664.2013.04.036 [14] CHENG Mingming, ZHANG Guoxin, MITRA N J, et al. Global contrast based salient region detection[C]// IEEE. International Conferene on Computer Vision and Pattern Recognition. Colorado: IEEE, 2011: 409 − 416. [15] 陈宁宁. 几种图像阈值分割算法的实现与比较[J]. 电脑知识与技术, 2011, 7(13): 3109 − 3111. CHEN Ningning. Achieve and comparison of image segmentation thresholding method [J]. Comput Knowl Technol, 2011, 7(13): 3109 − 3111. [16] 白雪冰, 许景涛, 郭景秋, 等. 基于局部二值拟合模型的板材表面节子与虫眼的图像分割[J]. 浙江农林大学学报, 2016, 33(2): 306 − 314. doi: 10.11833/j.issn.2095-0756.2016.02.017 BAI Xuebing, XU Jingtao, GUO Jingqiu, et al. Segmentation of wood surface knots and wormholes based on an improvedLBF Model [J]. J Zhejiang A&F Univ, 2016, 33(2): 306 − 314. doi: 10.11833/j.issn.2095-0756.2016.02.017 [17] BORJI J, CHENG Mingming, JIANG Huaizu, et al. Salient object detection: a benchmark [J]. IEEE Trans Image Process, 2012, 24(12): 5706 − 5722. -
-
链接本文:
http://zlxb.zafu.edu.cn/article/doi/10.11833/j.issn.2095-0756.20200318
点击查看大图
计量
- 文章访问数: 18
- 被引次数: 0
