Terrain Classification Research Articles

Terrain classification and rock abundance analysis at Utopia Planitia using Zhurong image data based on deep learning algorithms

The complexity of image scene information presents challenges for the trafficability assessment and path planning of Mars rovers. To ensure the operational safety of Mars rovers and extract terrain features from complex image scenes, this paper develops an end-to-end deep learning model, using the deep convolutional neural networks ResNet50 and DeepLabV3 + as the framework, with images from the Zhurong rover’s navigation camera as the training and test datasets. A deep learning model suitable for classification and segmentation of terrain in the Mars Utopia Planitia region has been established and applied to planetary geology research. The classification accuracy of model exceeds 83.90 % and segmentation accuracy exceeds 80 %. Subsequently, an analysis of 1309 raw images from the navigation camera yielded 203,744 individual estimates of rock abundance, the model evaluates the rock abundance in the Utopia Planitia region, where the Zhurong rover is located, at 10.94 %. The terrain classification model proposed in this study provides both engineering and scientific value for future rovers on the Utopia Planitia.

Road terrain recognition based on tire noise for autonomous vehicle

Effective road terrain recognition is crucial for enhancing the driving safety, passability, and comfort of autonomous vehicles. This study addresses the challenges of accurately identifying diverse road surfaces using deep learning in complex environments. We introduce a novel end-to-end Tire Noise Recognition Residual Network (TNResNet) integrated with a time-frequency attention module, designed to capture and leverage time-frequency information from tire noise signals for road terrain classification. Our method was evaluated on five distinct road types: asphalt, cement, grass, mud, and sand. The performance of TNResNet was rigorously compared against traditional machine learning techniques, including Decision Trees, K-Nearest Neighbors, and Support Vector Machines, as well as advanced deep learning models like Long Short-Term Memory and Convolutional Neural Networks. Experimental results demonstrate that TNResNet achieves superior classification accuracy of 99.48%, outperforming all comparative methods. This work not only establishes a robust framework for road terrain identification but also showcases the significant practical implications of TNResNet in the realm of autonomous vehicle navigation.

A terrain data collection sensor box towards a better analysis of terrains conditions

<p>Autonomous mobile robots are increasingly used across various applications, relying on multiple sensors for environmental awareness and efficient task execution. Given the unpredictability of human environments, versatility is crucial for these robots. Their performance is largely determined by how they perceive their surroundings. This paper introduces a machine learning (ML) approach focusing on land conditions to enhance a robot’s locomotion. The authors propose a method to classify terrains for data collection, involving the design of an apparatus to gather field data. This design is validated by correlating collected data with the output of a standard ML model for terrain classification. Experiments show that the data from this apparatus improves the accuracy of the ML classifier, highlighting the importance of including such data in the dataset.</p>

ANALYSING THE IMPACT OF URBAN GROWTH ON AGRICULTURAL LANDS USING SLEUTH MODEL AND GOOGLE EARTH ENGINE

In this study, it is aimed to determine the urban growth in the Selçuklu district of Konya, which is the study area with the SLEUTH model based on cellular automata, which is widely used in the modeling of urban growth and land use, and to examine the effect of urbanization on agricultural areas in the near future. In addition to the simulations carried out for the years 2030 and 2050 starting from 2015, which was determined as the last control year in the model, the simulation results of the year 2022 were compared with the terrain classes obtained from the Google Earth Engine (GEE) controlled classification of the 2022 Landsat satellite image. As a result of the creation of simulation models for the years 2030 and 2050, it was concluded that 10428.75-23747.49 hectares of agricultural land will be destroyed, respectively. The SLEUTH model has modeled a total of 56468.26 hectares of agricultural land for 2022. This corresponds to 95% of the classification result for 2022, which is an important factor in examining the accuracy of the model. This study, which aims to guide decision makers and planners, shows that the use of the SLEUTH model has strong implications for the planned examination of future land use.

Comparison of Deep Learning approaches in classification of lacial landforms

Glacial landforms, created by the continuous movements of glaciers over millennia, are crucial topics in geomorphological research. Their systematic analysis affords invaluable insights into past climatic oscillations and augments understanding of long-term climate change dynamics. The classification of these types of terrain traditionally depends on labor-intensive manual or semi-automated methods. However, the emergence of automated techniques driven by deep learning and neural networks holds promise for enhancing efficiency of terrain classification workflows. This study evaluated the effectiveness of Convolutional Neural Network (CNN) architectures, particularly Residual Neural Network (ResNet) and VGG in comparison with Vision Transformer (ViT) architecture in the glacial landform classification task. By using preprocessed input data from Digital Elevation Model (DEM) which covers regions such as the Lubawa Upland and Gardno-Leba Plain in Poland, as well as the Elise Glacier in Svalbard, Norway, comprehensive assessments of those methods were conducted. The final results highlight the unique ability of deep learning methods to accurately classify glacial landforms. Classification process presented in this study can be the efficient, repeatable and fast solution for automatic terrain classification.

Terrain Classification Based on Spatiotemporal Multihead Attention With Flexible Tactile Sensors Array

Terrain Classification Based on Spatiotemporal Multihead Attention With Flexible Tactile Sensors Array

Terrain analysis in forested areas: consequences of using DSM instead of DTM

Abstract. Relief mapping through dense tropical forest is a challenge, which can be met by processing radar images. Various global digital elevation models (DEMs) are available, either free or paid, the most widely used methods for producing them at global scale being photogrammetry and short wavelength radar interferometry. However, the resulting DEMs do not directly represent the ground surface but rather the canopy or something in between. Since most users need terrain models for a variety of applications in geoscience, water management etc. but use such canopy models for lack of anything better, it was relevant to assess the consequences of this inappropriate but unavoidable choice. This paper presents a study carried out in the Brazilian Amazon, where an airborne P-band radar interferometric DEM was used as a reference based on its accuracy assessed in previous studies. A well-established DEM (SRTM) and newer ones (Copernicus and Fabdem), as well as a DEM obtained from aerial X-bands radar data, were compared to the reference according to different criteria related to elevation (accuracy and a typical application, i.e., dam filling simulation) and to slope at different scales (accuracy and a typical application, i.e., terrain classification). The results highlighted the risks of using short wavelengths to represent the terrain and emphasized the importance of slope in different resolution scales.

Minimum data sampling requirements for accurate detection of terrain-induced gait alterations change with mobile sensor position

Human gait is a key biomarker for health, independence and quality of life. Advances in wearable inertial sensor technologies have paved the way for out-of-the-lab human gait analysis, which is important for the assessment of mobility and balance in natural environments and has applications in multiple fields from healthcare to urban planning. Automatic recognition of the environment where walking takes place is a prerequisite for successful characterisation of terrain-induced gait alterations. A key question which remains unexplored in the field is how minimum data requirements for high terrain classification accuracy change depending on the sensor placement on the body. To address this question, we evaluate the changes in performance of five canonical machine learning classifiers by varying several data sampling parameters including sampling rate, segment length, and sensor configuration. Our analysis on two independent datasets clearly demonstrate that a single inertial measurement unit is sufficient to recognise terrain-induced gait alterations, accuracy and minimum data requirements vary with the device position on the body, and choosing correct data sampling parameters for each position can improve classification accuracy up to 40% or reduce data size by 16 times. Our findings highlight the need for adaptive data collection and processing algorithms for resource-efficient computing on mobile devices.

Robot terrain classification based on improved Hilbert-Huang transform and long short-term memory network

Nowadays, with the rapid development of science and technology, the innovation and application of robot technology has become an important force to promote industrial development, which requires robots to have a high degree of autonomy and adaptability. Among them, terrain classification technology is one of the key technologies to achieve this goal. In order to improve the ground adaptability of robots in complex environments, this paper proposes a terrain classification algorithm based on improved Hilbert-Huang transform(HHT) combined with ensemble empirical mode decomposition(EEMD) and long short-term memory network(LSTM). Firstly, the signal data is processed by EEMD, and then the frequency domain features of the signal are extracted by Hilbert transform to expand the feature dimension. Finally, the features are learned and classified by the LSTM model, which effectively improves the classification accuracy. In this paper, we conducted sufficient experiments to compare the effects of EMD and EEMD and the effects of different neural network models, and verified the contribution of the Hilbert-Huang transform to improve the classification performance through ablation experiments, which proves the effectiveness and reliability of our proposed algorithm, and provided powerful technical support for the robot to adapt to the ground information in the complex environment.

Three-dimensional modeling of 500 kV transmission lines by airborne LiDAR

This study aims to enhance the accuracy of three-dimensional (3D) modeling in the planning, design, operation management, and risk assessment of 500 kV transmission lines. Addressing the limitations of traditional two-dimensional (2D) or low-precision 3D models, a new method based on airborne Light Detection and Ranging (LiDAR) technology is proposed. Initially, high-density, high-precision 3D point cloud data are collected using airborne LiDAR technology. This data is then processed using specialized software for noise removal, terrain classification extraction, and Geographic Information System (GIS) data integration. Subsequently, a Convolutional Neural Network (CNN) is employed to classify the 3D point cloud data to enhance the accuracy of complex structure recognition. Lastly, the point cloud data are converted into linear and planar entities using Computer-Aided Design (CAD) technology, achieving high-precision 3D reconstruction of the 500 kV transmission lines and their complex terrain environment. Experimental results show that the spatial resolution of the constructed model reaches 0.1 m per pixel, with overall geometric accuracy within ±5 cm. Compared to traditional aerial imagery and ground survey methods, the model based on airborne LiDAR technology achieves improvements of 30 % in detail richness and 35 % in terrain adaptability, with a 45 % increase in decision-making efficiency. These improvements significantly enhance the accuracy of line operations, safety assessments, and environmental impact analyses, providing solid data support for the modernization and intelligent development of the power industry.

A Robot Ground Medium Classification Algorithm Based on Feature Fusion and Adaptive Spatio-Temporal Cascade Networks

With technological advancements and scientific progress, mobile robots have found widespread applications across various fields. To enable robots to perform tasks safely and effectively in diverse and unknown environments, this paper proposes a ground medium classification algorithm for robots based on feature fusion and an adaptive spatio-temporal cascade network. Specifically, the original directional features in the dataset are first transformed into quaternion form. Then, spatio-temporal forward and reverse neighbors are identified using KD trees, and their connection strengths are evaluated via a kernel density estimation algorithm to determine the final set of neighbors. Subsequently, based on the connection strengths determined in the previous step, we perform noise reduction on the features using discrete wavelet transform. The noise-reduced features are then weighted and fused to generate a new feature representation.After feature fusion, the Adaptive Dynamic Convolutional Neural Network (ADC) proposed in this paper is cascaded with the Long Short-Term Memory (LSTM) network to further extract hybrid spatio-temporal feature information from the dataset, culminating in the final terrain classification. Experiments on the terrain type classification dataset demonstrate that our method achieves an average accuracy of 97.46% and an AUC of 99.80%, significantly outperforming other commonly used algorithms in the field. Furthermore, the effectiveness of each module in the proposed method is further demonstrated through ablation experiments.

Channel adaptive CVFCN using a new transfer method for PolSAR terrain classification

ABSTRACT This paper mainly addresses the lack of labelled data and insufficient data utilization in PolSAR image classification. We propose a channel adaptive Complex-Valued Fully Convolutional Networks based on Transfer learning (TF-CVFCN). To solve the problem of less labelled data, we borrowed from the transfer learning method, which can use a large amount of easily obtained relevant or even irrelevant data for pre-training and then use a small amount of source data for network training fine-tuning. Since PolSAR data can obtain different data information by using different decomposition methods, the current research trend is to process PolSAR data in the complex-valued containing more information. In our work, to cater to the proposed network model and make the network design more liberal, a new transfer learning method is proposed, which uses a new method to realize partial layer parameter transfer, and achieves the purpose of optimizing the traditional Transfer learning method. On this basis, we optimize the network structure to improve the classification accuracy of PolSAR data, and evaluate its performance on three datasets. Furthermore, we lighten the TF-CVFCN and optimize the relevant parameters using Competitive Multi-Objective Particle Swarm Optimizer, emphasizing the importance of this parameter optimization in large networks.

Shadow-aware terrain classification: advancing hyperspectral image sensing through generative adversarial networks and correlated sample synthesis

Shadow-aware terrain classification: advancing hyperspectral image sensing through generative adversarial networks and correlated sample synthesis

Gini Coefficient-Based Feature Learning for Unsupervised Cross-Domain Classification with Compact Polarimetric SAR Data

Remote sensing image classification usually needs many labeled samples so that the target nature can be fully described. For synthetic aperture radar (SAR) images, variations of the target scattering always happen to some extent due to the imaging geometry, weather conditions, and system parameters. Therefore, labeled samples in one image could not be suitable to represent the same target in other images. The domain distribution shift of different images reduces the reusability of the labeled samples. Thus, exploring cross-domain interpretation methods is of great potential for SAR images to improve the reuse rate of existing labels from historical images. In this study, an unsupervised cross-domain classification method is proposed that utilizes the Gini coefficient to rank the robust and stable polarimetric features in both the source and target domains (GRFST) such that an unsupervised domain adaptation (UDA) can be achieved. This method selects the optimal features from both the source and target domains to alleviate the domain distribution shift. Both fully polarimetric (FP) and compact polarimetric (CP) SAR features are explored for crop-domain terrain type classification. Specifically, the CP mode refers to the hybrid dual-pol mode with an arbitrary transmitting ellipse wave. This is the first attempt in the open literature to investigate the representing abilities of different CP modes for cross-domain terrain classification. Experiments are conducted from four aspects to demonstrate the performance of CP modes for cross-data, cross-scene, and cross-crop type classification. Results show that the GRFST-UDA method yields a classification accuracy of 2% to 12% higher than the traditional UDA methods. The degree of scene similarity has a certain impact on the accuracy of cross-domain crop classification. It was also found that when both the FP and circular CP SAR data are used, stable, promising results can be achieved.

Graphic Characterization of Terrain Using Information Computational Methods

As is well known, automated pattern recognition and satellite data analysis include a broad spectrum of pattern recognition. Image classification is a crucial remote sensing technology that facilitates the automatic understanding of massive amounts of data. Numerous researchers have put a great deal of effort into this field,and their results are impressive. In order to distinguish between diverse terrain classes, such as water, urban,rocky, vegetation, and barren, we are examining a variety of nature-inspired strategies in this article. the many methods we are using to do this, including the hybrid algorithms of Particle Swarm Optimization (PSO),Artificial Bee Colony (ABC) and Biogeography Based Optimization (BBO), Firefly and Firefly Algorithm, and Hybrid Cuckoo Search & Firefly Algorithm. These methods with a strong natural influence are crucial for applications involving distant sensing. These methods have shown good results showing increased efficiency and larger values of the Kappa coefficient, making them highly helpful for classifying aspects of the terrain. This research examines many methods of classifying satellite photos that draw inspiration from nature.

Multi-Step Unsupervised Domain Adaptation in Image and Feature Space for Synthetic Aperture Radar Image Terrain Classification

The significant differences in data domains between SAR images and the expensive and time-consuming process of data labeling pose significant challenges to terrain classification. Current terrain classification methodologies face challenges in addressing domain disparities and detecting uncommon terrain effectively. Based on Style Transformation and Domain Metrics (STDMs), we propose an unsupervised domain adaptive framework named STDM-UDA for terrain classification in this paper, which consists of two steps: image style transfer and domain adaptive segmentation. As a first step, image style transfer is performed within the image space to mitigate the differences in low-level features between SAR image domains. Subsequently, leveraging this process, the segmentation network extracts image features, employing domain metrics and adversarial training to enhance alignment between domain gaps in the semantic feature space. Finally, experiments conducted on several pairs of SAR images, each exhibiting varying degrees of differences in key imaging parameters such as source, resolution, band, and polarization, demonstrate the robustness of the proposed method. It achieves remarkably competitive classification accuracy, particularly for unlabeled, high-resolution broad scenes, effectively overcoming the domain gaps introduced by the diverse imaging parameters under studies.

ENHANCED TERRAIN RECOGNITION WITH DEEP LEARNING

Terrain recognition is vital for numerous real-world applications, from autonomous navigation to disaster management. Convolutional Neural Networks (CNNs) have emerged as potent tools for addressing terrain recognition challenges. In this paper, we propose an innovative approach to significantly improve terrain recognition accuracy using CNNs. We meticulously curate a dataset from Kaggle, comprising 1989 high-resolution images categorized into four terrain classes: Grassy, Sandy, Rocky, and Marshy. Our methodology revolves around the systematic design and implementation of deep learning techniques, primarily focusing on CNN architectures. Additionally, we contribute by training a CNN model tailored for classifying images into the four terrain classes. Leveraging the computational resources of Google Colab, we conduct extensive experimentation and analysis to evaluate the performance of our CNN-based terrain recognition system. Empirical results demonstrate substantial advancements in terrain recognition accuracy, underscoring the transformative role of CNNs in enhancing the efficiency and precision of terrain classification systems. Furthermore, we delve into the intricacies of our CNN model's architecture, exploring key design choices and optimization strategies. These insights deepen our understanding of CNN- based terrain recognition systems and provide valuable guidance for future research endeavors. Overall, our study highlights the practical relevance and transformative potential of CNNs in elevating terrain recognition accuracy. Keywords: Terrain recognition, Convolutional Neural Networks (CNNs), Image classification, Real-world applications, Dataset, Performance evaluation, Optimization strategies, Deep learning

CONVCAT: A NEW CLASSIFICATION APPROACH USING UC MERCED AND RESISC45 DATASETS

With advances in Earth observation systems, the importance of remote sensing data is increasing daily. These data are used in various fields ranging from image segmentation to terrain classification, from disaster impact assessment to climate change analysis. The use of remotely sensed images for terrain classification has been the subject of a number of studies. This study proposes a new method for terrain classification in the UC Merced Land Use Dataset and RESISC45 remote sensing images. This method is called ConvCat model, which is a combination of classical convolutional layer and CatBoost models. The performance of this model is measured in terms of accuracy, the Matthews Correlation Coefficient (MCC) and the Cohen's Kappa metrics. The results are compared with ensemble models (XGBoost, CatBoost), with ConvXGB, a combination of convolutional learning and XGBoost, and with ResNet50, one of the most widely used transfer learning models. The developed ConvCat model outperformed the other models, achieving an accuracy of 97.44% on the UC Merced data set and an accuracy of 96.89% on the Resisc45 data set. This study shows that our newly developed model provides the best results for the classification problem based on remote sensing images.

JPSSL: SAR Terrain Classification Based on Jigsaw Puzzles and FC-CRF

Effective features play an important role in synthetic aperture radar (SAR) image interpretation. However, since SAR images contain a variety of terrain types, it is not easy to extract effective features of different terrains from SAR images. Deep learning methods require a large amount of labeled data, but the difficulty of SAR image annotation limits the performance of deep learning models. SAR images have inevitable geometric distortion and coherence speckle noise, which makes it difficult to extract effective features from SAR images. If effective semantic context features cannot be learned for SAR images, the extracted features struggle to distinguish different terrain categories. Some existing terrain classification methods are very limited and can only be applied to some specified SAR images. To solve these problems, a jigsaw puzzle self-supervised learning (JPSSL) framework is proposed. The framework comprises a jigsaw puzzle pretext task and a terrain classification downstream task. In the pretext task, the information in the SAR image is learned by completing the SAR image jigsaw puzzle to extract effective features. The terrain classification downstream task is trained using only a small number of labeled data. Finally, fully connected conditional random field processing is performed to eliminate noise points and obtain a high-quality terrain classification result. Experimental results on three large-scene high-resolution SAR images confirm the effectiveness and generalization of our method. Compared with the supervised methods, the features learned in JPSSL are highly discriminative, and the JPSSL achieves good classification accuracy when using only a small amount of labeled data.

Terrain-Vechicle Traction Control Parameters for Tracked-Robot

Abstract: Here we have the systems of :- 1) Advancements in Vehicle Control :- This summary highlights two recent advancements in vehicle control technology which are the enhanced Handling for Passenger Cars: Researchers have developed a novel control system inspired by expert driver techniques. This system optimizes a car's handling during maneuvers requiring simultaneous acceleration and sharp turns. Simulation testing demonstrates significant performance improvements. Improved Unmanned Ground Vehicle (UGV) Operation on Unstructured Terrain: A new approach addresses the challenges of controlling UGVs on rough terrain. The solution combines sensor fusion for terrain characterization, a neural network for terrain classification, and an adaptive control system for real-time stability adjustments. This multi-pronged approach has been validated through successful experimentation on a UGV platform. 2) Enhancing UGV Control with Haptic Feedback :- A separate study explores the use of haptic feedback to improve UGV control in difficult environments. Traditional operator interfaces often lack sufficient sensory information, leading to reduced situational awareness and operational difficulties. This research proposes a haptic tablet interface that provides the operator with a tactile sense of the UGV's traction state. User studies confirm that this approach significantly reduces the time a UGV spends immobilized due to traction loss. These advancements collectively represent significant progress in vehicle control technology, with potential benefits for both manned and unmanned vehicles operating in challenging scenarios