Object-based Environment Research Articles

(Mal)Adaptive Learning After Switches Between Object-Based and Rule-Based Environments

In reinforcement-learning studies, the environment is typically object-based; that is, objects are predictive of a reward. Recently, studies also adopted rule-based environments in which stimulus dimensions are predictive of a reward. In the current study, we investigated how people learned (1) in an object-based environment, (2) following a switch to a rule-based environment, (3) following a switch to a different rule-based environment, and (4) following a switch back to an object-based environment. To do so, we administered a reinforcement-learning task comprising of four blocks with consecutively an object-based environment, a rule-based environment, another rule-based environment, and an object-based environment. Computational-modeling results suggest that people (1) initially adopt rule-based learning despite its suboptimal nature in an object-based environment, (2) learn rules after a switch to a rule-based environment, (3) experience interference from previously-learned rules following a switch to a different rule-based environment, and (4) learn objects after a final switch to an object-based environment. These results imply people have a hard time adjusting to switches between object-based and rule-based environments, although they do learn to do so.

A Review of Testing Object-Based Environment Perception for Safe Automated Driving

Safety assurance of automated driving systems must consider uncertain environment perception. This paper reviews literature addressing how perception testing is realized as part of safety assurance. The paper focuses on testing for verification and validation purposes at the interface between perception and planning, and structures the analysis along the three axes (1) test criteria and metrics, (2) test scenarios, and (3) reference data. Furthermore, the analyzed literature includes related safety standards, safety-independent perception algorithm benchmarking, and sensor modeling. It is found that the realization of safety-oriented perception testing remains an open issue since challenges concerning the three testing axes and their interdependencies currently do not appear to be sufficiently solved.

Automated detection of rock glaciers using deep learning and object-based image analysis

Rock glaciers are an important component of the cryosphere and are one of the most visible manifestations of permafrost. While the significance of rock glacier contribution to streamflow remains uncertain, the contribution is likely to be important for certain parts of the world. High-resolution remote sensing data has permitted the creation of rock glacier inventories for large regions. However, due to the spectral similarity between rock glaciers and the surrounding material, the creation of such inventories is typically conducted based on manual interpretation, which is both time consuming and subjective. Here, we present a novel method that combines deep learning (convolutional neural networks or CNNs) and object-based image analysis (OBIA) into one workflow based on freely available Sentinel-2 optical imagery (10 m spatial resolution), Sentinel-1 interferometric coherence data, and a digital elevation model (DEM). CNNs identify recurring patterns and textures and produce a prediction raster, or heatmap where each pixel indicates the probability that it belongs to a certain class (i.e. rock glacier) or not. By using OBIA we can segment the datasets and classify objects based on their heatmap value as well as morphological and spatial characteristics. We analysed two distinct catchments, the La Laguna catchment in the Chilean semi-arid Andes and the Poiqu catchment in the central Himalaya. In total, our method mapped 108 of the 120 rock glaciers across both catchments with a mean overestimation of 28%. Individual rock glacier polygons howevercontained false positives that are texturally similar, such as debris-flows, avalanche deposits, or fluvial material causing the user's accuracy to be moderate (63.9–68.9%) even if the producer's accuracy was higher (75.0–75.4%). We repeated our method on very-high-resolution Pléiades satellite imagery and a corresponding DEM (at 2 m resolution) for a subset of the Poiqu catchment to ascertain what difference image resolution makes. We found that working at a higher spatial resolution has little influence on the producer's accuracy (an increase of 1.0%), however the rock glaciers delineated were mapped with a greater user's accuracy (increase by 9.1% to 72.0%). By running all the processing within an object-based environment it was possible to both generate the deep learning heatmap and perform post-processing through image segmentation and object reshaping. Given the difficulties in differentiating rock glaciers using image spectra, deep learning combined with OBIA offers a promising method for automating the process of mapping rock glaciers over regional scales and lead to a reduction in the workload required in creating inventories.

Basic Procedures and Methodological Support of the “Object-Based Environment for Engineering Calculation in the MATLAB System” Course

This paper looks at methodological aspects of the «Object-based environment for engineering calculation in the MATLAB system» course offered to the second year students of Bauman Moscow State Technical University for mastering tools in the MATLAB environment, learning MATLAB language and object-oriented technology of working with classes and applications, designed for engineering calculations of ammunition characteristics. The paper presents the technology of organization of practical studies using resources of MATLAB Grader.

Multimodal Object-Based Environment Representation for Assistive Robotics

Autonomous robots are nowadays successfully used in industrial environments, where tasks follow predetermined plans and the world is a known (and closed) set of objects. The context of social robotics brings new challenges to the robot. First of all, the world is no longer closed. New objects can be introduced at any time, and it is now impossible to build an exaustive list of them nor having a precomputed set of descriptors. Moreover, natural interactions with a human being don’t follow any precomputed graph of sequences or grammar. To deal with the complexity of such an open world, a robot can no longer solely rely on its sensors data: a compact representation to comprehend its surrounding is needed. Our approach focuses on task independent environment representation where human-robot interactions are involved. We propose a global architecture bridging the gap between perception and semantic modalities through instances (physical realizations of semantic concepts). In this article, we describe a method for automatic generation of object-related ontology. Based on it, a practical formalization of the ill-defined notion of “context” is discussed. We then tackle human-robot interactions in our system through the description of user request processing. Finally, we illustrate the flow of our model on two showcases which demonstrate the validity of the approach.

Implementing an object-based multi-index protocol for mapping surface glacier facies from Chandra-Bhaga basin, Himalaya

Surface glacier facies are superficial expressions of a glacier that are distinguishable based on differing spectral and structural characteristics according to their age and inter-mixed impurities. Increasing bodies of literature suggest that the varying properties of surface glacier facies differentially influence the melt of the glacier, thus affecting the mass balance. Incorporating these variations into distributed mass balance modelling can improve the perceived accuracy of these models. However, detecting and subsequently mapping these facies with a high degree of accuracy is a necessary precursor to such complex modelling. The variations in the reflectance spectra of various glacier facies permit multiband imagery to exploit band ratios for their effective extraction. However, coarse and medium spatial resolution multispectral imagery can delimit the efficacy of band ratioing by muddling the minor spatial and spectral variations of a glacier. Very high-resolution imagery, on the other hand, creates distortions in the conventionally obtained information extracted through pixel-based classification. Therefore, robust and adaptable methods coupled with higher resolution data products are necessary to effectively map glacier facies. This study endeavours to identify and isolate glacier facies on two unnamed glaciers in the Chandra-Bhaga basin, Himalayas, using an established object-based multi-index protocol. Exploiting the very high resolution offered by WorldView-2 and its eight spectral bands, this study implements customized spectral index ratios via an object-based environment. Pixel-based supervised classification is also performed using three popular classifiers to comparatively gauge the classification accuracies. The object-based multi-index protocol delivered the highest overall accuracy of 86.67%. The Minimum Distance classifier yielded the lowest overall accuracy of 62.50%, whereas, the Mahalanobis Distance and Maximum Likelihood classifiers yielded overall accuracies of 77.50% and 70.84% respectively. The results outline the superiority of the object-based method for extraction of glacier facies. Forthcoming studies must refine the indices and test their applicability in wide ranging scenarios.

Explorative Study on Mapping Surface Facies of Selected Glaciers from Chandra Basin, Himalaya Using WorldView-2 Data

Mapping of surface glacier facies has been a part of several glaciological applications. The study of glacier facies in the Himalayas has gained momentum in the last decade owing to the implications imposed by these facies on the melt characteristics of the glaciers. Some of the most commonly reported surface facies in the Himalayas are snow, ice, ice mixed debris, and debris. The precision of the techniques used to extract glacier facies is of high importance, as the result of many cryospheric studies and economic reforms rely on it. An assessment of a customized semi-automated protocol against conventional and advanced mapping algorithms for mapping glacier surface facies is presented in this study. Customized spectral index ratios (SIRs) are developed for effective extraction of surface facies using thresholding in an object-based environment. This method was then tested on conventional and advanced classification algorithms for an evaluation of the mapping accuracy for five glaciers located in the Himalayas, using very high-resolution WorldView-2 imagery. The results indicate that the object-based image analysis (OBIA) based semi-automated SIR approach achieved a higher average overall accuracy of 87.33% (κ = 0.85) than the pixel-based image analysis (PBIA) approach. Among the conventional methods, the Maximum Likelihood performed the best, with an overall accuracy of 78.71% (κ = 0.75). The Constrained Energy Minimization, with an overall accuracy of 68.76% (κ = 0.63), was the best performer of the advanced algorithms. The advanced methods greatly underperformed in this study. The proposed SIRs show a promise in the mapping of minor features such as crevasses and in the discrimination between ice-mixed debris and debris. We have efficiently mapped surface glacier facies independently of short-wave infrared bands (SWIR). There is a scope for the transferability of the proposed SIRs and their performance in varying scenarios.

Spectral indices based object oriented classification for change detection using satellite data

Change detection, using remotely sensed data can be utilized in a diversified way such as, land use and cover analysis, forest or vegetation assessment, and flood monitoring. The aim of this study is to develop a methodology for change detection in highly urbanized areas, using time-series satellite imagery. This paper analyzes the effectiveness of the object oriented classification over unsupervised algorithms such as k-means for the purpose of change detection. The study area selected is, National Capital Territory of Delhi, which is a good representative of the urban agglomeration conditions in the Asian region. A time series of Landsat 5 (TM) and Landsat 8 (OLI) imageries for the duration of 1993–2014, have been acquired in order to represent the wide range of pattern variation. This study uses three spectral indices namely, Normalized Difference Built-up Index to characterize built-up area, Modified Normalized Difference Water Index to signify open water and Modified Soil Adjusted Vegetation Index to symbolize green vegetation. Further, this study employs object-based environment that separates similar pixels spatially and spectrally at different scales and assign information class to segmented objects. Results recommends the use of OBIA as a semi-automated tool for classification of remotely sensed satellite data. Moreover, using OBIA along with traditional spectral indices, for the classification, provides dimensionality reduction. Moreover, such a method produces classified map which have great correspondence with reality, reflects in high accuracies that were achieved for classes like built-up. The statistics indicated that the classification with object oriented paradigm achieved an overall accuracy of up to 90.1 %, which is far better as compared to automated unsupervised clustering algorithm, such as k-means.

Automated classification of debris-covered glaciers combining optical, SAR and topographic data in an object-based environment

Abstract Satellite imagery is increasingly used to monitor glacier area changes and create glacier inventories. Robust and efficient pixel-based band ratios have proven to be accurate for automatically delineating clean glacier ice, however such classifications are restricted on debris-covered ice due to its spectral similarity with surrounding terrain. Object-Based Image Analysis (OBIA) has emerged as a new analysis technique within remote sensing. It offers many advantages over pixel-based classification techniques due to the ability to work with multiple data sources and handle data contextually and hierarchically. By making use of OBIA capabilities we automatically classify clean ice and debris-covered ice in the challenging area surrounding Mount Manaslu in Nepal using optical (Landsat 8), topographic (void-filled SRTM) and SAR coherence (ALOS PALSAR) data. Clean ice was classified with a mean accuracy of 93% whilst debris-covered ice was classified with an accuracy of 83% when compared to manually corrected outlines, providing a total glacier accuracy of 91%. With further developments in the classification, steep tributary sections of ice could be contextually included, raising the accuracy to over 94%. One prominent advantage of OBIA is that it allows some post-processing and correction of the glacier outlines automatically, reducing the amount of manual correction needed. OBIA incorporating SAR coherence data is recommended for future mapping of debris-covered ice.

Detection and classification of mesoscale atmospheric phenomena above sea in SAR imagery

Satellite radar observations above sea are dedicated to small scale and mesoscale phenomena in the ocean and the marine boundary layer. What the Synthetic Aperture Radar (SAR) sensors see on the ocean surface is primarily the variation of the Bragg resonant waves providing signatures of local wind changes and imprints of atmospheric phenomena, and the areas of enhanced wave breaking associated with wave–current interaction for oceanographic processes like currents, fronts and eddies. SAR images contain information about the processes in the marine–atmosphere boundary layer (MABL), in particular when variation of wind speed and direction is significant. Several publications are available in the literature modeling and explaining each phenomenon separately. However, only a few focus on the identification of each phenomenon in an automatic or semi-automatic way. The objective of the present paper is to detect and to classify five atmospheric phenomena in SAR images: low wind areas, katabatic wind, island wakes, atmospheric gravity waves and atmospheric convective cells. An object-based environment was used to identify the phenomena and fuzzy logic used to classify them. Several features extracted for each phenomenon were used for feeding the classifier. The proposed method used 16 SAR images and accuracy was examined in another 16 SAR images using 457 object samples. The method proved to be successful with an overall accuracy of 80% and Kappa index of 0.75. The class of low wind areas negatively affected the overall accuracy due to overlapping feature values with the rest of the categories.

Land Cover Change Detection Using Object-Based Classification Technique: A Case Study Along The Kosi River, Bihar

Abstract. Land use/land cover (LULC) is dynamic in nature and can affect the ability of land to sustain human activities. The Indo-Gangetic plains of north Bihar in eastern India are prone to floods, which have a significant impact on land use / land cover, particularly agricultural lands and settlement areas. Satellite remote sensing techniques allow generating reliable and near-realtime information of LULC and have the potential to monitor these changes due to periodic flood. Automated methods such as object-based techniques have better potential to highlight changes through time series data analysis in comparison to pixel-based methods, since the former provides an opportunity to apply shape, context criteria in addition to spectral criteria to accurately characterise the changes. In this study, part of Kosi river flood plains in Supaul district, Bihar has been analysed to identify changes due to a flooding event in 2008. Object samples were collected from the post-flood image for a nearest neighbourhood (NN) classification in an object-based environment. Collection of sample were partially supported by the existing 2004–05 database. The feature space optimisation procedure was adopted to calculate an optimum feature combination (i.e. object property) that can provide highest classification accuracy. In the study, for classification of post-flood image, best class separation was obtained by using distance of 0.533 for 28 parameters out of 34. Results show that the Kosi flood has resulted in formation of sandy riverine areas.

Earth Observation-Based Dwelling Detection Approaches in a Highly Complex Refugee Camp Environment — A Comparative Study

For effective management of refugee camps or camps for internally displaced persons (IDPs) relief organizations need up-to-date information on the camp situation. In cases where detailed field assessments are not available, Earth observation (EO) data can provide important information to get a better overview about the general situation on the ground. In this study, different approaches for dwelling detection were tested using the example of a highly complex camp site in Somalia. On the basis of GeoEye-1 imagery, semi-automatic object-based and manual image analysis approaches were applied, compared and evaluated regarding their analysis results (absolute numbers, population estimation, spatial pattern), statistical correlations and production time. Although even the results of the visual image interpretation vary considerably between the interpreters, there is a similar pattern resulting from all methods, which shows same tendencies for dense and sparse populated areas. The statistical analyses revealed that all approaches have problems in the more complex areas, whereas there is a higher variance in manual interpretations with increasing complexity. The application of advanced rule sets in an object-based environment allowed a more consistent feature extraction in the area under investigation that can be obtained at a fraction of the time compared to visual image interpretation if large areas have to be observed.

A Semi-Automated Object-Based Approach for Landslide Detection Validated by Persistent Scatterer Interferometry Measures and Landslide Inventories

Geoinformation derived from Earth observation (EO) plays a key role for detecting, analyzing and monitoring landslides to assist hazard and risk analysis. Within the framework of the EC-GMES-FP7 project SAFER (Services and Applications For Emergency Response) a semi-automated object-based approach for landslide detection and classification has been developed. The method was applied to a case study in North-Western Italy using SPOT-5 imagery and a digital elevation model (DEM), including its derivatives slope, aspect, curvature and plan curvature. For the classification in the object-based environment spectral, spatial and morphological properties as well as context information were used. In a first step, landslides were classified on a coarse segmentation level to separate them from other features with similar spectral characteristics. Thereafter, the classification was refined on a finer segmentation level, where two categories of mass movements were differentiated: flow-like landslides and other landslide types. In total, an area of 3.77 km² was detected as landslide-affected area, 1.68 km² were classified as flow-like landslides and 2.09 km² as other landslide types. The outcomes were compared to and validated by pre-existing landslide inventory data (IFFI and PAI) and an interpretation of PSI (Persistent Scatterer Interferometry) measures derived from ERS1/2, ENVISAT ASAR and RADARSAT-1 data. The spatial overlap of the detected landslides and existing landslide inventories revealed 44.8% (IFFI) and 50.4% (PAI), respectively. About 32% of the polygons identified through OBIA are covered by persistent scatterers data.

Morphological Building/Shadow Index for Building Extraction From High-Resolution Imagery Over Urban Areas

Morphological building index (MBI) is a recently developed approach for automatic indication of buildings in high-resolution imagery. However, MBI is subject to commission errors due to the similar characteristics between buildings, bare soil and roads. Furthermore, omission errors occur in dark and heterogeneous roofs. In this study, a systematic framework for building extraction from high-resolution imagery is proposed, aiming to alleviate both commission and omission errors for the original MBI algorithm. The improvements include three aspects: 1) a morphological shadow index (MSI) is proposed to detect shadows that are used as a spatial constraint of buildings; 2) a dual-threshold filtering is proposed to integrate the information of MBI and MSI; 3) the proposed framework is implemented in an object-based environment, where a geometrical index and a vegetation index are then used to remove noise from narrow roads and bright vegetation. The proposed framework was validated on an Ikonos image of Washington DC Mall with 1-m resolution and an 8-channel WorldView-2 image of Hangzhou, east of China, with 2-m resolution. By comparison with the ground truth references, it was shown that our method achieved over 90% overall accuracy for discrimination between buildings and backgrounds for both datasets. In the comparative study, it was revealed that the proposed method improved the original MBI significantly. Furthermore, the proposed method was more accurate than the support vector machine interpretation with the differential morphological profiles (DMP) and multiscale urban complexity index (MUCI).

Object-oriented analysis of multi-temporal panchromatic images for creation of historical landslide inventories

Object-oriented analysis (OOA) has been demonstrated to produce more accurate results than pixel-based image processing. Studies carried out by previous researchers have shown how landslide inventories can be prepared from multispectral satellite images using OOA. However, panchromatic images are frequently the only data available after a landslide event. Furthermore, preparation of historical inventories relies on the analysis of satellite images and aerial photographs acquired over past few decades that are also mostly only available in black and white. In such cases the methodology developed using multispectral data cannot be used directly due to limited spectral information, in particular in near-infrared bands. In this paper we present a new methodology that addresses some of these issues. Using high resolution panchromatic images from Cartosat-1 (2.5 m) and IRS-1D (5.8 m), and a 10 m gridded DTM extracted from Cartosat-1, we developed a new approach which uses change detection techniques and a global contextual criteria in an object-based environment to detect and classify landslides into five different types. Continuous time series images from 1998 to 2006 were used to prepare annual landslide inventories in a highly rugged Himalayan terrain. The maximum and minimum detection percentages achieved for all landslides are 96.7% and 71.5%, respectively, with corresponding quality percentages of 88.1% and 55.3%, respectively. However, the lack of spectral information proved to be a hurdle resulting in a high branching factor that indicates that further work is required to eliminate false positives. Nevertheless, the method was able to create much needed historical landslide inventories, which are critical for landslide hazard and risk assessment studies.

A Conceptual Model for Multi-Temporal Landscape Monitoring in an Object-Based Environment

Remote sensing plays a critical role in contemporary monitoring programs, but our strategies for processing these data using automated procedures are not always reliable. In particular, the task of separating real from spurious changes remains problematic, especially in an object-based environment where differential errors in classification quality, spatial registration, scene illumination, resolution, and object delineation have forced some operators to adopt labor-intensive visual-interpretation strategies, or employ manual interaction on an object-by-object basis. In this paper, we present an updated summary of our new disturbance-inventory approach to land-cover monitoring that combines object-based classification and change-detection strategies with boundary-conditioning routines designed to maximize the spatial and thematic integrity of the finished products. With this approach, the final maps are only altered in regions of confirmed change, and spurious gaps, slivers, stretches, and encroachments are avoided. The approach constitutes an innovative, efficient, and transparent framework that can handle all the basic landscape dynamics, including feature appearance, disappearance, succession, expansion, and shrinkage, without the need for manual editing.

Simplified Distributed Programming with Micro Objects

Developing large-scale distributed applications can be a daunting task. object-based environments have attempted to alleviate problems by providing distributed objects that look like local objects. We advocate that this approach has actually only made matters worse, as the developer needs to be aware of many intricate internal details in order to adequately handle partial failures. The result is an increase of application complexity. We present an alternative in which distribution transparency is lessened in favor of clearer semantics. In particular, we argue that a developer should always be offered the unambiguous semantics of local objects, and that distribution comes from copying those objects to where they are needed. We claim that it is often sufficient to provide only small, immutable objects, along with facilities to group objects into clusters.

Texture and Scale in Object-Based Analysis of Subdecimeter Resolution Unmanned Aerial Vehicle (UAV) Imagery

Imagery acquired with unmanned aerial vehicles (UAVs) has great potential for incorporation into natural resource monitoring protocols due to their ability to be deployed quickly and repeatedly and to fly at low altitudes. While the imagery may have high spatial resolution, the spectral resolution is low when lightweight off-the-shelf digital cameras are used, and the inclusion of texture measures can potentially increase the classification accuracy. Texture measures have been used widely in pixel-based image analysis, but their use in an object-based environment has not been well documented. Our objectives were to determine the most suitable texture measures and the optimal image analysis scale for differentiating rangeland vegetation using UAV imagery segmented at multiple scales. A decision tree was used to determine the optimal texture features for each segmentation scale. Results indicated the following: (1) The error rate of the decision tree was lower; (2) prediction success was higher; (3) class separability was greater; and (4) overall accuracy was higher (high 90% range) at coarser segmentation scales. The inclusion of texture measures increased classification accuracies at nearly all segmentation scales, and entropy was the texture measure with the highest score in most decision trees. The results demonstrate that UAVs are viable platforms for rangeland monitoring and that the drawbacks of low-cost off-the-shelf digital cameras can be overcome by including texture measures and using object-based image analysis which is highly suitable for very high resolution imagery.

Object-based approaches to change analysis and thematic map update: challenges and limitations

small, spurious polygons created by the inconsistent delineation of persistent change features appearing in consecutive coregistered images. The issue represents a serious methodological challenge that can limit the visual and structural quality of the finished map product if not adequately addressed. A critical analysis of annual land cover maps generated by updating and backdating object-based reference maps in a western Alberta study area revealed that sliver objects made up between 3% and 12% of the total area of change, and between 63% and 72% of the total number of change objects, despite high thematic accuracies. The results highlight the emerging need for a methodological framework designed to handle the spatial challenges posed by change analysis in an object-based environment.

Object-based sound synthesis for virtual environments-using musical acoustics

We show how to construct an object-based environment for sound generation, whose objects can be individually synthesized and which can interact with each other through modelling of a potential interaction topology. We show how this interaction topology can be mode dynamic and time-varying. We also discuss how we envision an object-based environment that integrates geometric, radiometric, and intrinsic/extrinsic acoustic properties. We finally illustrate our first results towards modelling of complex sound generation systems.