Cloud detection in remote sensing images is a crucial task in various applications,such as meteorological disaster prediction and earth resource exploration,which require accurate cloud identi¯cation.This work pr...Cloud detection in remote sensing images is a crucial task in various applications,such as meteorological disaster prediction and earth resource exploration,which require accurate cloud identi¯cation.This work proposes a cloud detection model based on the Cloud Detection neural Network(CDNet),incorporating a fusion mechanism of channel and spatial attention.Depthwise separable convolution is adopted to achieve a lightweight network model and enhance the e±ciency of network training and detection.In addition,the Convolutional Block Attention Module(CBAM)is integrated into the network to train the cloud detection model with attention features in channel and spatial dimensions.Experiments were conducted on Landsat 8 imagery to validate the proposed improved CDNet.Averaged over all testing images,the overall accuracy(OA),mean Pixel Accuracy(mPA),Kappa coe±cient and Mean Intersection over Union(MIoU)of improved CDNet were 96.38%,81.18%,96.05%,and 84.69%,respectively.Those results were better than the original CDNet and DeeplabV3+.Experiment results show that the improved CDNet is e®ective and robust for cloud detection in remote sensing images.展开更多
针对传统云检测算法对噪声敏感、提取轮廓不精确等问题,提出一种结合语义分割神经网络结构U-Net和后处理算法TTA(test time augmentation)的云检测方法,实现了高精度云检测的同时还很好地保留了云边缘轮廓。首先,利用U-Net网络的U型结...针对传统云检测算法对噪声敏感、提取轮廓不精确等问题,提出一种结合语义分割神经网络结构U-Net和后处理算法TTA(test time augmentation)的云检测方法,实现了高精度云检测的同时还很好地保留了云边缘轮廓。首先,利用U-Net网络的U型结构挖掘云覆盖区域像元高级、低级特征;其次,通过TTA增强待云检测的影像特征,提升模型鲁棒性。实验结果表明,结合U-Net结构和TTA的云检测精度达到93.2%,高于其他传统算法约5%,解决了经典算法对噪声敏感的缺点,提高了仅使用U-Net时的云检测精度。展开更多
For a vision measurement system consisted of laser-CCD scanning sensors, an algorithm is proposed to extract and recognize the target object contour. Firstly, the two-dimensional（2D） point cloud that is output by th...For a vision measurement system consisted of laser-CCD scanning sensors, an algorithm is proposed to extract and recognize the target object contour. Firstly, the two-dimensional（2D） point cloud that is output by the integrated laser sensor is transformed into a binary image. Secondly, the potential target object contours are segmented and extracted based on the connected domain labeling and adaptive corner detection. Then, the target object contour is recognized by improved Hu invariant moments and BP neural network classifier. Finally, we extract the point data of the target object contour through the reverse transformation from a binary image to a 2D point cloud. The experimental results show that the average recognition rate is 98.5% and the average recognition time is 0.18 s per frame. This algorithm realizes the real-time tracking of the target object in the complex background and the condition of multi-moving objects.展开更多
基金supported by the National Natural Science Foundation of China (61973164,62373192).
文摘Cloud detection in remote sensing images is a crucial task in various applications,such as meteorological disaster prediction and earth resource exploration,which require accurate cloud identi¯cation.This work proposes a cloud detection model based on the Cloud Detection neural Network(CDNet),incorporating a fusion mechanism of channel and spatial attention.Depthwise separable convolution is adopted to achieve a lightweight network model and enhance the e±ciency of network training and detection.In addition,the Convolutional Block Attention Module(CBAM)is integrated into the network to train the cloud detection model with attention features in channel and spatial dimensions.Experiments were conducted on Landsat 8 imagery to validate the proposed improved CDNet.Averaged over all testing images,the overall accuracy(OA),mean Pixel Accuracy(mPA),Kappa coe±cient and Mean Intersection over Union(MIoU)of improved CDNet were 96.38%,81.18%,96.05%,and 84.69%,respectively.Those results were better than the original CDNet and DeeplabV3+.Experiment results show that the improved CDNet is e®ective and robust for cloud detection in remote sensing images.
文摘针对传统云检测算法对噪声敏感、提取轮廓不精确等问题,提出一种结合语义分割神经网络结构U-Net和后处理算法TTA(test time augmentation)的云检测方法,实现了高精度云检测的同时还很好地保留了云边缘轮廓。首先,利用U-Net网络的U型结构挖掘云覆盖区域像元高级、低级特征;其次,通过TTA增强待云检测的影像特征,提升模型鲁棒性。实验结果表明,结合U-Net结构和TTA的云检测精度达到93.2%,高于其他传统算法约5%,解决了经典算法对噪声敏感的缺点,提高了仅使用U-Net时的云检测精度。
文摘For a vision measurement system consisted of laser-CCD scanning sensors, an algorithm is proposed to extract and recognize the target object contour. Firstly, the two-dimensional（2D） point cloud that is output by the integrated laser sensor is transformed into a binary image. Secondly, the potential target object contours are segmented and extracted based on the connected domain labeling and adaptive corner detection. Then, the target object contour is recognized by improved Hu invariant moments and BP neural network classifier. Finally, we extract the point data of the target object contour through the reverse transformation from a binary image to a 2D point cloud. The experimental results show that the average recognition rate is 98.5% and the average recognition time is 0.18 s per frame. This algorithm realizes the real-time tracking of the target object in the complex background and the condition of multi-moving objects.
