| 潘鹏亮, 张方梅, 尹 健, 刘红敏, 周顺玉, 智亚楠.菊方翅网蝽雌雄成虫图像识别技术的初步研究[J].植物保护,2017,43(3):70-75. |
| 菊方翅网蝽雌雄成虫图像识别技术的初步研究 |
| Preliminary studies on image recognition technology for female and male adults of Corythucha marmorata (Uhler) (Hemiptera: Tingidae) |
| 投稿时间：2016-09-18 修订日期：2016-12-04 |
| DOI： |
| 中文关键词: 菊方翅网蝽 形态特征 图像处理 雌雄鉴定 自动识别 |
| 英文关键词:Corythucha marmorata morphological feature image processing sex indentification automatic identification |
| 基金项目:信阳农林学院校内青年教师基金(201401015),豫南植物有害生物绿色防控院士工作站项目(YZ201506) |
|
| 摘要点击次数: 1133 |
| 全文下载次数: 1596 |
| 中文摘要: |
| 菊方翅网蝽主要为害菊科植物,是特种经济作物菊芋的潜在性害虫,对此害虫的形态识别和种群动态预测非常重要。昆虫几何形态特征的提取和分析是进行其自动识别、远程鉴定、种群监测和预测预报的基础。本文利用图像采集与处理技术,通过BugShape软件对其外部几何形态特征进行了自动和手动获取与分析。结果表明,菊方翅网蝽周长和紧凑度两个参数在不同分辨率下差异显著,雌虫球状性在不同分辨率下也差异显著。雌雄之间周长、长轴长度和短轴长度差异不显著,而偏心率、紧凑度、球状性、叶状性和圆形度差异均达到显著水平。雌雄个体的原始判别正确率为92%~93%,交叉判别正确率为91%~93%。使用所有10个参数进行判别分析时,判别的正确率与分辨率呈正相关关系。手动测量结果表明,菊方翅网蝽雌雄成虫体长分别为(1.86±0.07)mm和(1.80±0.07)mm,中胸宽度分别为(0.73±0.02)mm和(0.69±0.03)mm,腹部宽度分别为(1.04±0.05)mm和(0.84±0.07)mm。利用手动测量数据进行雌雄个体判别分析的正确率达到98%~99%。 |
| 英文摘要: |
| The chrysanthemum lace bug Corythucha marmorata (Uhler) is one of the potential pests of Jerusalem artichoke Helianthus tuberosus. It is very important to know its morphological characteristics and forecast the dynamics of its population. The extraction and analysis of insect geometrical features are fundamental researches for the insect automatic recognition, long distance identification, population monitoring and forecasting. In this study, the geometrical features of the lace bug were obtained automatically and manually by using the software BugShape v1.0. The results showed that the significant difference emerged in terms of two parameters (perimeter and compactness) and sphericity of females under different scanning resolutions. The parameters (eccentricity, compactness, sphericity, lobation and circularity) were significantly different between males and females, but not in the parameters of perimeter, long axis and shirt axis. 92%-93% samples were correctly classified by original validation, and 91%-93% correctly classified by cross validation under different resolutions by using the above five parameters. However, the discriminant rates were positively related with the scanning resolutions when all parameters were extracted by the software. The results from the data by manual showed that the body lengths of females and males were (1.86±0.07)mm and (1.80±0.07)mm, respectively; the mesothorax width was （0.73±0.02）mm and (0.69±0.03)mm, and the abdomen width was (1.04±0.05)mm and (0.84±0.07)mm. 98%-99% samples were correctly classified by cross validation by using the three parameters and the ratios between them. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
