Search

The tunnel environment is characterized by insufficient ambient light, obstructed view, and complex inner lining construction conditions. These factors frequently result in limited anti-interference capability, reduced recognition accuracy, and suboptimal segmentation results for defect extraction. We propose a deep network model utilizing an encoder–decoder framework that integrates Transformer and convolution for comprehensive defect extraction. The proposed model utilizes an encoder that integrates a hierarchical Transformer backbone with an efficient attention mechanism to fully explore complete information at multi-scale granularities. In the decoder, multi-scale information is initially aggregated using a Multi-Layer Perceptron (MLP) module. Additionally, the Stacking Filters with Atrous Convolutions (SFAC) module are implemented to enhance the perception of the complete defect scope. Furthermore, a Boundary-aware Attention Module (BAM) is implemented to enhance edge information to improve the detection of defects. With this well-designed decoder, the multi-scale information from the encoder can be fully aggregated and exploited for complete defect detection. Experimental findings illustrate the effectiveness of our proposed approach in addressing tunnel lining defects within the image dataset. The outcomes reveal that our proposed network achieves an accuracy (Acc) of 94.4% and a mean intersection over union (mIoU) of 78.14%. Compared to state-of-the-art segmentation networks, our model improves the accuracy of tunnel lining defect extraction, showcasing enhanced extraction effectiveness and anti-interference capability, thus meeting the engineering requirements for defect detection in complex environments of tunnels.

The success of transformer networks in the natural language processing and 2D vision domains has encouraged the adaptation of transformers to 3D computer vision tasks. However, most of the existing approaches employ standard backpropagation (SBP). SBP requires the storage of model activations on a forward pass for use during the backward pass, making their memory complexity linearly proportional to model depth, hence, inefficient. Furthermore, most 3D point transformers use the classic QKV matrix multiplication design which comes with a memory bottleneck. To address these issues, we propose a memory-efficient point transformer that makes use of reversible functions and linearized self-attention to minimize SBP and transformer memory complexities, respectively. Additionally, rather than the usual UNet architectural design for segmentation, we adopt a ∇-shaped design to capture multi-size/resolution feature representations towards a finely detailed segmentation. Experimental results on benchmark datasets (Toronto3D, DALES, and CSPC) from different sensor platforms (vehicle, aerial, and backpack) show that our approach uses less than half the number of model parameters compared to its SBP counterpart. It also takes more than twice the input sequence and uses less than half the memory compared to most of the traditional approaches. Additionally, our use of a ∇-shaped architectural design yielded about a 5% increment in model performance over the U-shaped approach. Overall, the proposed PReFormer attained competitive performance compared to the state-of-the-art.

Urban scene-level 3D point cloud labeling is a very laborious and expensive task compared to images. Conversely however, image processing techniques, deep learning or otherwise are more established and mature. Thus, in a multi-source data environment, the labeling of a point cloud scene via an automated image process as an initial step, followed by a manual human verification process is an effective way to save man hours and cost. With the above as the goal, this study presents a simple but robust spatio-spectral feature representation approach. In this approach, a class-aware band selection and reduction (CBSR) technique is developed for optimal hyperspectral feature representation. A double-branched convolutional Gaussian Bernoulli deep belief network (CGBDBN) is then used for hierarchical spatial feature extraction from LiDAR-derived data and the CBSR data. Using stacked ensemble learning, spatio-spectral features are generated from the two feature streams via a fusion rule and then classified — the results of which are used in labeling a raw 3D LiDAR point cloud through projection. To evaluate this study, extensive experiments were conducted on the IEEE 2018 Houston dataset — the only publicly available dataset with both hyperspectral image (HSI) and 3D point cloud covering the same area — for urban scene classification. The results indicate that the developed CBSR attained comparatively competitive results with state-of-the-art approaches, thus making it a robust spectral feature representation technique. Also, the weight-sharing property, probabilistic modeling, and hierarchical nature of CGBDBN gives our approach the ability to capture high-level contextual features. Furthermore, compared to the spatial- or spectral-only features, the generated spatio-spectral features are more discriminative and significantly aided in improving the proposed model’s efficacy. Overall, the proposed approach, based on the evaluation metrics, is a robust and effective approach for both coarse- and fine-grained raw LiDAR point cloud labeling tasks.

Accurate individual tree segmentation is an important basis for the subsequent calculation and analysis of forestry parameters. However, rasterized canopy height model based methods often suffer from 3-D information loss due to the interpolation operation. Therefore, this article proposes an individual tree segmentation method based on the marker-controlled watershed algorithm and 3-D spatial distribution analysis from airborne LiDAR point clouds. First, based on the potential tree apices derived from the local maxima filtering, the marker-controlled watershed segmentation algorithm is conducted to obtain the coarse point clusters. Then, within the principal component analysis defined local coordinate reference framework, a multidirectional 3-D spatial profile analysis is performed on each point cluster to refine the potential tree apex positions. Finally, the refined potential tree apex positions are used as a prior of K-means clustering to achieve the coarse-to-fine individual tree segmentation. Comparative experiments were conducted on the public NEWFOR dataset to evaluate the proposed method. Results indicate that the proposed method is efficient and robust for segmenting individual trees.