Please wait a minute...
Journal of Zhejiang University (Agriculture and Life Sciences)  2017, Vol. 43 Issue (2): 262-272    DOI: 10.3785/j.issn.1008-9209.2016.04.113
Agricultural engineering     
 Evaluation on formation rate of Pleurotus eryngii primordium under different humidity conditions by computer vision
ZHOU Jun, DING Wenjie*, ZHU Xuejun, CAO Junyi, NIU Xueming
(School of Mechanical Engineering, Ningxia University, Yinchuan 750021, China)
Download: HTML (   PDF(3767KB)
Export: BibTeX | EndNote (RIS)       Supporting Info

Abstract  Humidity is one of significant factors affecting the quantity of Pleurotus eryngii primordium. Artificial statistics are necessary to count the number of primordium, since the model for prediction of the formation rate of primordium has not been developed. In this paper, computer vision based on statistics was applied to develop a formation rate model for primordium. To solve the problem of statistics on primordium, image preprocessing and gray recognition template extraction were firstly studied. The number of primordium was accounted on the basis of primordium size. However, recognition rate was low because of the similarity between primordium and background. Second, combined with the gray image matrix of primordium, a characteristic- genetic- screening method based on size and shape of primordium was proposed to extract the morphological characteristics of primordium seed, and a feature library of primordium seeds was built to display the characteristic data information. Then, the large data analysis was carried out on the morphological database based on genetic idea, and 12 seeds were acquired. A primordium quantity neural network prediction model was established based on back-propagation neural network in which matching quantity of primordium seeds was considered as input, the actual quantity of primordium as output. Primordium statistics was completed and verified, with accuracy up to 94.79%. According to the statistics on the primordium under different relative humidity conditions, the formation rate model of primordium was established. It is found that computer vision based statistical method for primordium can be used to evaluate the formation rate of primordium under different humidity conditions.

Published: 25 March 2017
CLC:  TP 391.4  
  S 646  
Fund:   National Natural Science Foundation of China (NO. 61263007), the National Science and Technology Support Program of China (No. 2013BAD16B04)
Corresponding Authors: DING Wenjie (http://orcid.org/0000-0001-5608-6016), E-mail: dwjnet@zju.edu.cn   
Cite this article:

ZHOU Jun, DING Wenjie, ZHU Xuejun, CAO Junyi, NIU Xueming.  Evaluation on formation rate of Pleurotus eryngii primordium under different humidity conditions by computer vision. Journal of Zhejiang University (Agriculture and Life Sciences), 2017, 43(2): 262-272.

URL:

http://www.zjujournals.com/agr/10.3785/j.issn.1008-9209.2016.04.113     OR     http://www.zjujournals.com/agr/Y2017/V43/I2/262


基于机器视觉的不同湿度下杏鲍菇原基形成速率评估(英文)

作为生长发育中的关键影响因子之一,湿度变化对控制杏鲍菇原基形成数量有重要的生产意义。然而,目前统计原基数量仍以人工为主且尚未建立起相应的原基形成速率模型。因此,本文采用以机器视觉为基础的原基数量统计方法来建立原基形成速率模型。为解决原基数量统计问题,首先对原基图像预处理、灰度识别模板提取等进行研究,采用以原基尺寸为依据的识别模板对原基数量进行识别统计,然而识别率较低;进而结合原基灰度图像矩阵表现形式,提出了基于原基尺寸和形状的“遗传-特征-筛选”的方法提取原基形态特征种子,并建立原基种子形态特征库,以便直观显示种子特征数据信息;接着采用基于遗传思想的原基种子挖掘方法对原基种子形态特征库进行大数据分析,得到12个适用于原基形态特征提取的种子。借助反向传播神经网络,以种子匹配原基数量为输入、实际原基数量为输出建立了原基数量神经网络预测模型,实现了原基数量的统计。验证结果表明,原基数量统计准确率达到94.79%。根据不同相对湿度下的原基数量统计结果,建立了原基形成速率变化模型。试验表明,基于机器视觉的原基数量统计方法能够对不同湿度下的原基形成速率进行评估。
 
No related articles found!