High-Resolution Remote Sensing Image Classification Method Based on Convolutional Neural Network and Restricted Conditional Random Field

While machine learning techniques have been increasingly applied to land cover classification problems, these techniques have not focused on separating exposed bare rock from soil covered areas. Therefore, we built a convolutional neural network (CNN) to differentiate exposed bare rock (rock) from soil cover (other). We made a training dataset by mapping exposed rock at eight test sites across the Sierra Nevada Mountains (California, USA) using USDA’s 0.6 m National Aerial Inventory Program (NAIP) orthoimagery. These areas were then used to train and test the CNN. The resulting machine learning approach classifies bare rock in NAIP orthoimagery with a 0.95 F 1 score. Comparatively, the classical OBIA approach gives only a 0.84 F 1 score. This is an improvement over existing land cover maps, which underestimate rock by almost 90%. The resulting CNN approach is likely scalable but dependent on high-quality imagery and high-performance algorithms using representative training sets informed by expert mapping. As image quality and quantity continue to increase globally, machine learning models that incorporate high-quality training data informed by geologic, topographic, or other topical maps may be applied to more effectively identify exposed rock in large image collections.

Section: Discussionmentioning

confidence: 56%

Section: Potential Classification Modelsmentioning

confidence: 99%

Where’s the Rock: Using Convolutional Neural Networks to Improve Land Cover Classification

Petliak¹,

Cerovski-Darriau

Zaliva

et al. 2019

“…This process leads to some ground objects being excessively enlarged or reduced. Furthermore, if the different parts of ground objects that are shadowed or not shadowed are processed in the same manner, the CRF result will contain more errors [31]. In our previous work, we proposed a method called the restricted conditional random field (RCRF) that can handle the above situation [31].…”

Section: Introductionmentioning

confidence: 99%

“…Furthermore, if the different parts of ground objects that are shadowed or not shadowed are processed in the same manner, the CRF result will contain more errors [31]. In our previous work, we proposed a method called the restricted conditional random field (RCRF) that can handle the above situation [31]. Unfortunately, the RCRF requires the introduction of samples to control its iteration termination and produce an output integrated image.…”

Section: Introductionmentioning

confidence: 99%

An End-to-End and Localized Post-Processing Method for Correcting High-Resolution Remote Sensing Classification Result Images

Pan

Zhao

2020

Self Cite

Since the result images obtained by deep semantic segmentation neural networks are usually not perfect, especially at object borders, the conditional random field (CRF) method is frequently utilized in the result post-processing stage to obtain the corrected classification result image. The CRF method has achieved many successes in the field of computer vision, but when it is applied to remote sensing images, overcorrection phenomena may occur. This paper proposes an end-to-end and localized post-processing method (ELP) to correct the result images of high-resolution remote sensing image classification methods. ELP has two advantages. (1) End-to-end evaluation: ELP can identify which locations of the result image are highly suspected of having errors without requiring samples. This characteristic allows ELP to be adapted to an end-to-end classification process. (2) Localization: Based on the suspect areas, ELP limits the CRF analysis and update area to a small range and controls the iteration termination condition. This characteristic avoids the overcorrections caused by the global processing of the CRF. In the experiments, ELP is used to correct the classification results obtained by various deep semantic segmentation neural networks. Compared with traditional methods, the proposed method more effectively corrects the classification result and improves classification accuracy.

“…Furthermore, domain-specific knowledge (e.g., spatial dependency [32] and multiscaling) related to remote sensing and the geosciences has not, or only partially, been considered and embedded in these models. Although fully connected CRFs [33] may be used to encode such domain knowledge, only limited related studies [34,35] have been reported in the field of remote sensing.…”

Section: Introductionmentioning

confidence: 99%

Deep Residual Autoencoder with Multiscaling for Semantic Segmentation of Land-Use Images

2019

Semantic segmentation is a fundamental means of extracting information from remotely sensed images at the pixel level. Deep learning has enabled considerable improvements in efficiency and accuracy of semantic segmentation of general images. Typical models range from benchmarks such as fully convolutional networks, U-Net, Micro-Net, and dilated residual networks to the more recently developed DeepLab 3+. However, many of these models were originally developed for segmentation of general or medical images and videos, and are not directly relevant to remotely sensed images. The studies of deep learning for semantic segmentation of remotely sensed images are limited. This paper presents a novel flexible autoencoder-based architecture of deep learning that makes extensive use of residual learning and multiscaling for robust semantic segmentation of remotely sensed land-use images. In this architecture, a deep residual autoencoder is generalized to a fully convolutional network in which residual connections are implemented within and between all encoding and decoding layers. Compared with the concatenated shortcuts in U-Net, these residual connections reduce the number of trainable parameters and improve the learning efficiency by enabling extensive backpropagation of errors. In addition, resizing or atrous spatial pyramid pooling (ASPP) can be leveraged to capture multiscale information from the input images to enhance the robustness to scale variations. The residual learning and multiscaling strategies improve the trained model’s generalizability, as demonstrated in the semantic segmentation of land-use types in two real-world datasets of remotely sensed images. Compared with U-Net, the proposed method improves the Jaccard index (JI) or the mean intersection over union (MIoU) by 4-11% in the training phase and by 3-9% in the validation and testing phases. With its flexible deep learning architecture, the proposed approach can be easily applied for and transferred to semantic segmentation of land-use variables and other surface variables of remotely sensed images.