Up-to-date 3D building models are important for many applications. Airborne very high resolution (VHR) images often acquired annually give an opportunity to create an up-to-date 3D model. Building segmentation is often the first and utmost step. Convolutional neural networks (CNNs) draw lots of attention in interpreting VHR images as they can learn very effective features for very complex scenes. This paper employs Mask R-CNN to address two problems in building segmentation: detecting different scales of building and segmenting buildings to have accurately segmented edges. Mask R-CNN starts from feature pyramid network (FPN) to create different scales of semantically rich features. FPN is integrated with region proposal network (RPN) to generate objects with various scales with the corresponding optimal scale of features. The features with high and low levels of information are further used for better object classification of small objects and for mask prediction of edges. The method is tested on ISPRS benchmark dataset by comparing results with the fully convolutional networks (FCN), which merge high and low level features by a skip-layer to create a single feature for semantic segmentation. The results show that Mask R-CNN outperforms FCN with around 15% in detecting objects, especially in detecting small objects. Moreover, Mask R-CNN has much better results in edge region than FCN. The results also show that choosing the range of anchor scales in Mask R-CNN is a critical factor in segmenting different scale of objects. This paper provides an insight into how a good anchor scale for different dataset should be chosen.

Surface-related multiple elimination (SRME) is a solid and effective approach for primary estimation. However, due to the imperfections in data and method (e.g. coarsely-sampled dataset and balancing effect of adaptive subtraction) multiple energy leakage is commonly seen in the results of SRME-predicted primaries. Assuming that the primaries and multiples do not correlate locally in the time-space domain, we are able to extract the leaked multiples from the initially estimated primaries using local primary-and-multiple orthogonalization. The proposed framework consists of two steps: an initial primary/multiple estimation step and a multiple-leakage extraction step. The initial step corresponds to SRME, which produces the initial estimated primary and multiple models. The second step is based on local primary-and-multiple orthogonalization to retrieve the leaked multiples, which can be seen as a remedy for correcting the initial estimated primary and multiple models. Thus, we can obtain a better primary output which has much less leaked multiple energy. We demonstrate a good performance of our proposed framework on both synthetic and field data, where it repairs the leakage of standard adaptive subtraction.

Semantic segmentation, especially for buildings, from the very high resolution (VHR) airborne images is an important task in urban mapping applications. Nowadays, the deep learning has significantly improved and applied in computer vision applications. Fully Convolutional Networks (FCN) is one of the tops voted method due to their good performance and high computational efficiency. However, the state-of-art results of deep nets depend on the training on large-scale benchmark datasets. Unfortunately, the benchmarks of VHR images are limited and have less generalization capability to another area of interest. As existing high precision base maps are easily available and objects are not changed dramatically in an urban area, the map information can be used to label images for training samples. Apart from object changes between maps and images due to time differences, the maps often cannot perfectly match with images. In this study, the main mislabeling sources are considered and addressed by utilizing stereo images, such as relief displacement, different representation between the base map and the image, and occlusion areas in the image. These free training samples are then fed to a pre-trained FCN. To find the better result, we applied fine-tuning with different learning rates and freezing different layers. We further improved the results by introducing atrous convolution. By using free training samples, we achieve a promising building classification with 85.6% overall accuracy and 83.77% F1 score, while the result from ISPRS benchmark by using manual labels has 92.02% overall accuracy and 84.06% F1 score, due to the building complexities in our study area.

As full waveform inversion (FWI) is a non-unique and typically ill-posed inversion problem, it needs proper regularization. To optimally include geologic information, we propose full waveform inversion with directional total variation as an L1-norm regularization. We design the directional total variation (TV) operator based on the local dip field, derived from the seismic image, instead of ignoring the local structural direction and only using horizontal/vertical gradients in traditional TV. With a synthetic example with complex structures based on the Marmousi model, we demonstrate that the proposed method is much more effective compared to both FWI without regularization and FWI with conventional TV regularization.