1.西南林业大学机械与交通学院，云南昆明 650224
扫 描 看 全 文
张赛,王应彪,杨谭等.基于改进LeNet-5模型的木材表面典型缺陷识别方法研究[J].木材科学与技术,2021,35(06):31-37.
ZHANG Sai,WANG Ying-biao,YANG Tan,et al.Detecting Method of the Wood Surface Defects Based on Modified LeNet-5 Model[J].Chinese Journal of Wood Science and Technology,2021,35(06):31-37.
针对传统木材缺陷识别方法效率低、精度不高及泛化能力差等问题，对传统LeNet-5模型进行改进：通过分别增加卷积层和池化层的层数至4层，以增加网络深度；采用批量归一化算法，以解决内部协变量位移过拟合的问题；改用Leaky Relu函数作为激活函数，并加入稀疏分类交叉熵作为损失函数，使用Adam作为优化器，来优化网络模型。应用改进LeNet-5模型对辐射松木材常见缺陷（结疤、裂痕）及无缺陷样本集进行识别试验，结果表明：相对于传统LeNet-5模型以及VGG19、AlexNet、ResNet-50三种经典模型，改进LeNet-5模型的训练集准确率最高为99.87%、验证集为99.43%，运算时间缩短，木材缺陷识别精度和效率提高。
In order to improve efficiency, accuracy and commonization, the traditional LeNet-5 model was modified for detecting wood surface defects by increasing the number of convolutional layers and the pooling layers to 4 so that the network depth was increased. The batch normalization algorithm was used to solve internal co-vitiate shift displacement over-fitting problem. Leaky Relu function was used as the activation function, then sparse categorical cross entropy was added as the loss function, as well as Adam was used as the optimizer to enhance the network model. The comparison test was carried out on the samples of ,Pinus radiata, with and without classical defects(knots and cracks). The results showed that the modified LeNet-5 model performed best comparing with VGG19, AlexNet and ResNet-50 classic models. The accuracy rate was 99.87%; the verification set was 99.43%; the operation time was reduced, therefore the accuracy and efficiency of detecting wood surface defects were improved.
木材缺陷检测改进LeNet-5模型深度学习卷积神经网络
wood defect detectionmodified LeNet-5 modeldeep learningconvolution neural network
范佳楠, 刘英, 杨雨图, 等. 机器视觉在木材缺陷检测领域应用研究进展[J]. 世界林业研究, 2020,33(3):32-37.
FAN J N，LIU Y，YANG Y T，et al. Research progress in the application of machine vision to wood defect detection[J]. World Forestry Research, 2020,33(3):32-37.
丁伟彪, 王宝金. 木材缺陷检测技术研究概况与发展趋势[J]. 林业机械与木工设备, 2019,47(1):4-8.
DING W B, WANG B J. Research and development trend of wood defect detection technology[J]. Forestry Machinery & Woodworking Equipment, 2019,47(1):4-8.
赵鹏, 赵匀, 陈广胜. 基于3D扫描技术的木材缺陷定量化分析[J]. 农业工程学报, 2017,33(7):171-176.
ZHAO P，ZHAO Y，CHEN G S. Quantitative analysis of wood defect based on 3D scanning technique[J].Transactions of the Chinese Society of Agricultural Engineering，2017,33(7):171-176.
王立扬,张瑜,沈群,薛勇. 基于改进型LeNet-5的苹果自动分级方法[J].中国农机化学报,2020,41(7):105-110.
WANG L Y, ZHANG Y, SHEN Q, et al. Automatic detecting and grading method of apples based on improved LeNet-5[J]. Journal of Chinese Agricultural Mechanization, 2020,41(7):105-110.
萧嘉奕,马艳,周竹,方益明.基于Faster R-CNN模型的木材表面节子缺陷检测[J].木材工业,2020,34(2):45-48.
XIAO J Y, MA Y, ZHOU Z, et al. Detection of timber surface knots based on faster R-CNN[J]. Chinese Journal of Wood Science and Technology, 2020, 34(2):45-48.
刘英, 周晓林, 胡忠康, 等. 基于优化卷积神经网络的木材缺陷检测[J]. 林业工程学报, 2019,4(1):115-120.
LIU Y, ZHOU X L, HU Z K, et al. Wood defect recognition based on optimized convolution neural network algorithm[J]. Journal of Forestry Engineering, 2019, 4(1):115-120.
李若尘, 朱悠翔, 孙卫民, 等. 基于深度学习的木材缺陷图像的识别与定位[J]. 数据采集与处理, 2020,35(3):494-505.
LI R C, ZHU Y X, SUN W M, et al. Recognition and localization of wood defect image based on deep learning[J]. Journal of Data Acquisition and Processing, 2020,35(3):494-505.
TING H, YING L, YABIN Y, et al. Application of deep convolutional neural network on feature extraction and detection of wood defects[J]. Measurement, DOI:10.1016/j.measurement.2019.107357http://dx.doi.org/10.1016/j.measurement.2019.107357.
肖雨晴,杨慧敏,王柯欣,易强.卷积神经网络在木材缺陷检测应用中的研究进展[J].木材科学与技术,2021,35(3):12-18.
XIAO Y Q, YANG H M, WANG K X, et al. Research progress of convolutional neural networks in the application of wood defect detection[J]. Chinese Journal of Wood Science and Technology,2021,35(3):12-18.
Kumar A, Gandhi C P, Zhou Y, et al. Improved deep convolution neural network (CNN) for the identification of defects in the centrifugal pump using acoustic images[J]. Applied Acoustics, DOI:10.1016/j.apacoust.2020.107399http://dx.doi.org/10.1016/j.apacoust.2020.107399.
Khan A, Sohail A, Zahoora U, et al. A survey of the recent architectures of deep convolutional neural networks[J]. Artificial Intelligence Review, DOI:10.1007/s10462-020-09825-6http://dx.doi.org/10.1007/s10462-020-09825-6.
Krizhevsky A, Sutskever I, Hinton G. ImageNet classification with deep convolutional neural networks: NIPS[J]. Communications of the ACM, 2017: DOI:10.1145/3065386.
Simonyan K, Zisserman A. Very Deep Convolutional Networks for Large-Scale Image Recognition[J]. Computer Science, 2014. DOI: https://ui.adsabs.harvard.edu/abs/2014arXiv1409.1556Shttp://dx.doi.org/https://ui.adsabs.harvard.edu/abs/2014arXiv1409.1556S.
He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]. USA: IEEE, 2016.
Lecun Y, Bottou L. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998,86(11):2278-2324.
张力超,马蓉,张垚鑫.改进的LeNet-5模型在苹果图像识别中的应用[J].计算机工程与设计,2018,39(11):3570-3575.
ZHANG L C, MA R, ZHANG Y X. Application of improved LeNet-5 model in Apple image recognition[J]. Computer Engineering and Design, 2018, 39(11): 3570-3575.
Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift[J]. JMLR.org, 2015.
相关作者
相关机构
京公网安备11010802024621
微信公众号