Despite recent advances of deep Convolutional Neural Networks (CNNs) in various computer vision tasks, their potential for classification of multispectral remote sensing images has not been thoroughly explored. In particular, the applications of deep CNNs using optical remote sensing data have focused on the classification of very high-resolution aerial and satellite data, owing to the similarity of these data to the large datasets in computer vision. Accordingly, this study presents a detailed investigation of state-of-the-art deep learning tools for classification of complex wetland classes using multispectral RapidEye optical imagery. Specifically, we examine the capacity of seven well-known deep convnets, namely DenseNet121, InceptionV3, VGG16, VGG19, Xception, ResNet50, and InceptionResNetV2, for wetland mapping in Canada. In addition, the classification results obtained from deep CNNs are compared with those based on conventional machine learning tools, including Random Forest and Support Vector Machine, to further evaluate the efficiency of the former to classify wetlands. The results illustrate that the full-training of convnets using five spectral bands outperforms the other strategies for all convnets. InceptionResNetV2, ResNet50, and Xception are distinguished as the top three convnets, providing state-of-the-art classification accuracies of 96.17%, 94.81%, and 93.57%, respectively. The classification accuracies obtained using Support Vector Machine (SVM) and Random Forest (RF) are 74.89% and 76.08%, respectively, considerably inferior relative to CNNs. Importantly, InceptionResNetV2 is consistently found to be superior compared to all other convnets, suggesting the integration of Inception and ResNet modules is an efficient architecture for classifying complex remote sensing scenes such as wetlands.

The synergistic use of spatial features with spectral properties of satellite images enhances thematic land cover information, which is of great significance for complex land cover mapping. Incorporating spatial features within the classification scheme have been mainly carried out by applying just low-level features, which have shown improvement in the classification result. By contrast, the application of high-level spatial features for classification of satellite imagery has been underrepresented. This study aims to address the lack of high-level features by proposing a classification framework based on convolutional neural network (CNN) to learn deep spatial features for wetland mapping using optical remote sensing data. Designing a fully trained new convolutional network is infeasible due to the limited amount of training data in most remote sensing studies. Thus, we applied fine tuning of a pre-existing CNN. Specifically, AlexNet was used for this purpose. The classification results obtained by the deep CNN were compared with those based on well-known ensemble classifiers, namely random forest (RF), to evaluate the efficiency of CNN. Experimental results demonstrated that CNN was superior to RF for complex wetland mapping even by incorporating the small number of input features (i.e., three features) for CNN compared to RF (i.e., eight features). The proposed classification scheme is the first attempt, investigating the potential of fine-tuning pre-existing CNN, for land cover mapping. It also serves as a baseline framework to facilitate further scientific research using the latest state-of-art machine learning tools for processing remote sensing data.

Despite recent research on the potential of dual- (DP) and full-polarimetry (FP) Synthetic Aperture Radar (SAR) data for crop mapping, the capability of compact polarimetry (CP) SAR data has not yet been thoroughly investigated. This is of particular concern, given the availability of such data from RADARSAT Constellation Mission (RCM) shortly. Previous studies have illustrated potential for accurate crop mapping using DP and FP SAR features, yet what contribution each feature makes to the model accuracy is not well investigated. Accordingly, this study examined the potential of the early- to mid-season (i.e., May to July) RADARSAT-2 SAR images for crop mapping in an agricultural region in Manitoba, Canada. Various classification scenarios were defined based on the extracted features from FP SAR data, as well as simulated DP and CP SAR data at two different noise floors. Both overall and individual class accuracies were compared for multi-temporal, multi-polarization SAR data using the pixel- and object-based random forest (RF) classification schemes. The late July C-band SAR observation was the most useful data for crop mapping, but the accuracy of single-date image classification was insufficient. Polarimetric decomposition features extracted from CP and FP SAR data produced relatively equal or slightly better classification accuracies compared to the SAR backscattering intensity features. The RF variable importance analysis revealed features that were sensitive to depolarization due to the volume scattering are the most important FP and CP SAR data. Synergistic use of all features resulted in a marginal improvement in overall classification accuracies, given that several extracted features were highly correlated. A reduction of highly correlated features based on integrating the Spearman correlation coefficient and the RF variable importance analyses boosted the accuracy of crop classification. In particular, overall accuracies of 88.23%, 82.12%, and 77.35% were achieved using the optimized features of FP, CP, and DP SAR data, respectively, using the object-based RF algorithm.

A twofold architecture based on GNSS multipath environment prediction and detection is presented in a context of loosely coupled and tightly coupled IMU/GNSS integration for navigation in urban areas. A signal quality monitoring group of techniques is applied for a platform self‐contained effort to detect and exclude multipath‐contaminated GNSS signals. Additionally, the sensor integration Kalman filter stochastic model is adjusted on‐the‐fly based on a GNSS multipath environment map. The map is populated by crowdsourcing and contains the spatial distribution of average carrier‐to‐noise‐density ratio measurements, linked to the probability of non‐line‐of‐sight, multipath‐contaminated, diffracted, and attenuated signal reception. To address the map availability issue, a random forest machine learning model is developed to propagate the map to the city areas not directly surveyed by the mapping fleet based on open‐access geographic data. The architecture performance is evaluated in the automotive scenario showing 13‐17% accuracy improvement compared to a conventional Kalman filter.

Automatic target detection with complicated shapes in high spatial resolution images is an ongoing challenge in remote sensing image processing. This is because most methods use spectral or texture information, which are not sufficient for detecting complex shapes. In this paper, a new detection framework, based on Spatial Pyramid Matching (SPM) and Locality- constraint Linear Coding (LLC), is proposed to solve this problem, and exemplified using airplane shapes. The process starts with partitioning the image into sub-regions and generating a unique histogram for local features of each sub-region. Then, linear Support Vector Machines (SVMs) are used to detect objects based on a pyramid-matching kernel, which analyses the descriptors inside patches in different resolution. In order to generate the histogram, first a point feature detector (e.g. SIFT) is applied on the patches, and then a quantization process is used to select local features. In this step, the k-mean method is used in conjunction with the locality-constrained linear coding method. The LLC forces the coefficient matrix in the quantization process to be local and sparse as well. As a result, the speed of the method improves around 24 times in comparison to using sparse coding for quantization. Quantitative analysis also shows improvement in comparison to just using k-mean, but the accuracy in comparison to using sparse coding is similar. Rotation and shift of the desired object has no effect on the obtained results. The speed and accuracy of this algorithm for high spatial resolution images make it capable for use in real-world applications.