An approach to identification of linear continuous-time system is studied with modulating functions. Based on wavelet analysis theory, the multi-resolution modulating functions are designed, and the corresponding filt...An approach to identification of linear continuous-time system is studied with modulating functions. Based on wavelet analysis theory, the multi-resolution modulating functions are designed, and the corresponding filters have been analyzed. Using linear modulating filters, we can obtain an identification model that is parameterized directly in continuous-time model parameters. By applying the results from discrete-time model identification to the obtained identification model, a continuous-time estimation method is developed. Considering the accuracy of parameter estimates, an instrumental variable (Ⅳ) method is proposed, and the design of modulating integral filter is discussed. The relationship between the accuracy of identification and the parameter of modulating filter is investigated, and some points about designing Gaussian wavelet modulating function are outlined. Finally, a simulation study is also included to verify the theoretical results.展开更多
针对真实环境下采集的病害图像中存在的大量噪声和复杂背景干扰,导致作物病害识别准确性和泛化性低的问题,该研究提出基于自适应BayesShrink和频-空特征融合的作物病害识别方法(adaptive BayesShrink and frequencyspatial domain featu...针对真实环境下采集的病害图像中存在的大量噪声和复杂背景干扰,导致作物病害识别准确性和泛化性低的问题,该研究提出基于自适应BayesShrink和频-空特征融合的作物病害识别方法(adaptive BayesShrink and frequencyspatial domain features fusion, AFSF-DCT)。首先,设计了自适应BayesShrink算法(Ad-BayesShrink)以减少噪声干扰,同时保留更多细节,降低识别模型提取病害特征的难度。然后提出基于频-空特征融合和动态交叉自注意机制的作物病害识别模型(crop leaf disease identification model based on frequency-spatial features fusion and dynamic cross-self-attention,FSF-DCT)。为实现全面的频-空特征映射,设计了基于离散小波变换(discrete wavelet transform,DWT)和倒残差结构(bneck)的频-空特征映射(DWT-Bneck)分支以捕获多尺度病害特征。频域分支设计了基于2D DWT的特征映射模块(2D DWT-based frequency-features decomposition module, DWFD)以捕获病害细节和纹理,用于补充空间域特征在全局信息表达上的不足。空间域分支在bneck中引入CBAM(convolutional block attention module)和Dynamic Shift Max激活函数以实现全面的空间特征映射。最后设计了动态交叉自注意特征融合模块(multi-scale features fusion network based on dynamic cross-self-attention, MDCS-DF)融合频-空特征并增强模型对病害特征的关注。结果表明,Ad-BayesShrink获得了35.78的最高峰值信噪比,优于VisuShrink和SUREShrink。FSF-DCT在自建数据集和2个开源数据集(PlantVillage和AI challenger 2018)上分别获得了99.20%、99.90%和90.75%的识别精度,且具有较小的参数量(7.48 M)和浮点运算数(4.62 G),优于当前大部分的主流识别模型。AFSF-DCT可为复杂背景下的作物叶片病害的快速精准检测提供模型参考。展开更多
基金This project was supported by China Postdoctoral Science Foundation (2003034466)Scientific Research Fund of Hunan Provincial Education Department (02B032).
文摘An approach to identification of linear continuous-time system is studied with modulating functions. Based on wavelet analysis theory, the multi-resolution modulating functions are designed, and the corresponding filters have been analyzed. Using linear modulating filters, we can obtain an identification model that is parameterized directly in continuous-time model parameters. By applying the results from discrete-time model identification to the obtained identification model, a continuous-time estimation method is developed. Considering the accuracy of parameter estimates, an instrumental variable (Ⅳ) method is proposed, and the design of modulating integral filter is discussed. The relationship between the accuracy of identification and the parameter of modulating filter is investigated, and some points about designing Gaussian wavelet modulating function are outlined. Finally, a simulation study is also included to verify the theoretical results.
文摘针对真实环境下采集的病害图像中存在的大量噪声和复杂背景干扰,导致作物病害识别准确性和泛化性低的问题,该研究提出基于自适应BayesShrink和频-空特征融合的作物病害识别方法(adaptive BayesShrink and frequencyspatial domain features fusion, AFSF-DCT)。首先,设计了自适应BayesShrink算法(Ad-BayesShrink)以减少噪声干扰,同时保留更多细节,降低识别模型提取病害特征的难度。然后提出基于频-空特征融合和动态交叉自注意机制的作物病害识别模型(crop leaf disease identification model based on frequency-spatial features fusion and dynamic cross-self-attention,FSF-DCT)。为实现全面的频-空特征映射,设计了基于离散小波变换(discrete wavelet transform,DWT)和倒残差结构(bneck)的频-空特征映射(DWT-Bneck)分支以捕获多尺度病害特征。频域分支设计了基于2D DWT的特征映射模块(2D DWT-based frequency-features decomposition module, DWFD)以捕获病害细节和纹理,用于补充空间域特征在全局信息表达上的不足。空间域分支在bneck中引入CBAM(convolutional block attention module)和Dynamic Shift Max激活函数以实现全面的空间特征映射。最后设计了动态交叉自注意特征融合模块(multi-scale features fusion network based on dynamic cross-self-attention, MDCS-DF)融合频-空特征并增强模型对病害特征的关注。结果表明,Ad-BayesShrink获得了35.78的最高峰值信噪比,优于VisuShrink和SUREShrink。FSF-DCT在自建数据集和2个开源数据集(PlantVillage和AI challenger 2018)上分别获得了99.20%、99.90%和90.75%的识别精度,且具有较小的参数量(7.48 M)和浮点运算数(4.62 G),优于当前大部分的主流识别模型。AFSF-DCT可为复杂背景下的作物叶片病害的快速精准检测提供模型参考。
