An innovative framework integrating object-based image analysis and deep learning for urban landuse/landcover classification using WorldView-3 images

In recent years, deep learning-based image classification has become widespread, especially in remote sensing applications, due to its automatic and strong feature extraction capability. However, as deep learning methods operate on rectangular-shaped image patches, they cannot accurately extract objects’ boundaries, especially in complex urban settings. As a result, combining deep learning and object-based image analysis (OBIA) has become a new avenue in remote sensing studies. Therefore, the purpose of this study is to assess the current (DL-OBIA) integration frameworks for remote sensing image classification and then innovate a new framework to optimize this integration in terms of accuracy and computational efficiency. On the other hand, the proposed framework is validated for land cover mapping and road extraction applications. The proposed approach combines convolutional neural networks (CNN) with OBIA based on joint optimization of segmentation parameters and deep feature extraction. A Bayesian technique was used to find the best parameters for the multiresolution segmentation (MRS) algorithm while the CNN model learned the image features at different layers, achieving joint optimization. The proposed classification framework achieved the best accuracy, with 0.96 OA, 0.95 Kappa, and 0.96 mIoU in the training area and 0.97 OA, 0.96 Kappa, and 0.97 mIoU in the test area, outperforming several benchmark methods including Patch CNN, Center OCNN, Random OCNN, and Decision Fusion. The analysis of CNN variants within the proposed classification workflow showed that the HybridSN model achieved the best results compared to 2D and 3D CNNs. The 3D CNN layers and combining 3D and 2D CNN layers (HybridSN) yielded slightly better accuracies than the 2D CNN layers regarding geometric fidelity, object boundary extraction, and separation of adjacent objects. The Bayesian optimization could find comparable optimal MRS parameters for the training and test areas, with excellent quality measured by AFI (0.046, −0.037) and QR (0.945, 0.932). In the proposed framework, higher accuracies could be obtained with larger patch sizes (e.g., 9 × 9 compared to 3 × 3). Moreover, the proposed framework is computationally efficient, with the longest training being fewer than 25 s considering all the subprocesses and a single training epoch. The validation process shows the reliability of the proposed model for accurate classification from high-resolution satellite images. On the other hand, the framework’s validation on road extraction achieved an overall accuracy of 0.8637. The presented framework provides a new insight into OBIA-deep learning integration through the application of a Bayesian optimization technique, which offers a mutual optimization between segmentation and classification processes. Furthermore, it facilitates the intelligent selection of segmentation parameters. This can have a positive impact on future studies focusing on OBIA-deep learning integration by enhancing classification accuracy, preserving precise object boundaries, and reducing the number of iterations. This makes it more suitable for studies that require precise ground objects and accurate classification, such as building extraction, road detection, cloud detection, tree classification, shoreline, and damage assessment, which typically use very high-resolution imagery sources.

Text FK 2024 41 - Declaration Form.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (604kB) Text FK 2024 41 - Full Text.pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (8MB) Text FK 2024 41.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (8MB) Official URL or Download Paper: http://ethesis.upm.edu.my/id/eprint/19006

Actions (login required)

View Item