计算机应用 ›› 2021, Vol. 41 ›› Issue (3): 911-916.DOI: 10.11772/j.issn.1001-9081.2020060864

所属专题: 前沿与综合应用

• 前沿与综合应用 • 上一篇    

结合图像增强和卷积神经网络的小麦不完善粒识别

贺杰安1, 吴晓红1, 何小海1, 胡建蓉2, 卿粼波1   

  1. 1. 四川大学 电子信息学院, 成都 610065;
    2. 中储粮成都储藏研究院有限公司, 成都 610091
  • 收稿日期:2020-06-22 修回日期:2020-10-19 出版日期:2021-03-10 发布日期:2021-01-15
  • 通讯作者: 贺杰安
  • 作者简介:贺杰安(1996-),男,四川泸州人,硕士研究生,主要研究方向:图像处理、模式识别、人工智能;吴晓红(1970-),女,四川遂宁人,副教授,博士,主要研究方向:图像处理、模式识别、电子与通信系统;何小海(1964-),四川绵阳人,教授,博士,主要研究方向:图像处理、模式识别、图像通信、机器视觉、智能系统;胡建蓉(1984-),女,四川成都人,实验师,主要研究方向:粮油检测;卿粼波(1982-),男,四川简阳人,副教授,博士,主要研究方向:图像处理、模式识别、视频通信。
  • 基金资助:
    四川省科技计划项目(2018HH0143);四川省教育厅项目(18ZB0355)。

Imperfect wheat kernel recognition combined with image enhancement and conventional neural network

HE Jiean1, WU Xiaohong1, HE Xiaohai1, HU Jianrong2, QIN Linbo1   

  1. 1. College of Electronics and Information Engineering, Sichuan University, Chengdu Sichuan 610065, China;
    2. Sinograin Chengdu Reserves Institute Company Limited, Chengdu Sichuan 610091, China
  • Received:2020-06-22 Revised:2020-10-19 Online:2021-03-10 Published:2021-01-15
  • Supported by:
    This work is partially supported by the Science and Technology Program of Sichuan Province (2018HH0143), the Sichuan Provincial Department of Education Project (18ZB0355).

摘要: 针对实际应用场景下,小麦籽粒图像背景单一以及小麦不完善粒的不完善特征大多是局部特征而大部分图像特征与正常粒无异的特点,提出一种基于细节的图像增强(IE)的小麦不完善粒识别方法。首先,使用交替最小化算法约束原图在水平方向和竖直方向的L0范数来平滑原图作为基础图层,并用原图减去基础图层得到图像的细节层;然后,突出细节层后将其与基础图层叠加以增强图像;最后,将增强后的图像作为卷积神经网络(CNN)的输入,使用加入了批正则化(BN)层的CNN对图像进行识别。分别以经典分类网络LeNet-5、ResNet-34、VGG-16和在其中添加BN层的这些网络作为分类网络,增强前后的图像作为输入来进行分类实验,并以测试集准确率评估性能。实验结果表明,三个经典分类网络均在添加了BN层后而使用相同输入时的测试集准确率提高了5个百分点,在使用细节增强后的图像作为输入时三个网络的测试集准确率提高了1个百分点,以上二者联合使用时三个网络均获得超过7个百分点的测试集准确率提升。

关键词: 小麦不完善粒识别, 卷积神经网络, L0平滑, 图像增强, 批正则化, 分类

Abstract: In the practical application scenario, the wheat kernel image background is single, and the imperfect characteristics of wheat imperfect grains are mostly local features while most of the image features are not different from normal grains. In order to solve the problems, an imperfect wheat kernel recognition method based on detail Image Enhancement (IE) was proposed. Firstly, the alternate minimization algorithm was used to constrain the L0 norms of the original image in the horizontal and vertical directions to smooth the original image as the base layer, and the original image was subtracted from the base layer to obtain the detail layer of the image. Then, the detail layer was delighted and superimposed with the base layer to enhance the image. Finally, the enhanced image was used as the input of the Convolutional Neural Network (CNN), and the CNN with Batch Normalization (BN) layer was used for recognition of the image. The classic classification networks LeNet-5, ResNet-34, VGG-16 and these networks with the BN layer were used as classification networks, and the images before and after enhancement were used as input to carry out classification experiments, and the accuracy of the test set was used to evaluate the performance. Experimental results show that by adding the BN layer and using the same input, all three classic classification networks have the accuracy of the test set increased by 5 percentage points, and when using the images with enhanced detail as input, the three networks have the accuracy of the test set increased by 1 percentage point, and when the above two are used together, all the three networks obtain the accuracy of the test set improved by more than 7 percentage points.

Key words: imperfect wheat kernel recognition, Conventional Neural Network (CNN), L0 smoothing, image enhancement, Batch Normalization (BN), classification

中图分类号: