Supervised learning on graphs of spatio-temporal similarity in satellite image sequences
نویسندگان
چکیده
High resolution satellite image sequences are multidimensional signals composed of spatiotemporal patterns associated to numerous and various phenomena. Bayesian methods have been previously proposed in [8] to code the information contained in satellite image sequences in a graph representation using Bayesian methods. Based on such a representation, this paper further presents a supervised learning methodology of semantics associated to spatio-temporal patterns occurring in satellite image sequences. It enables the recognition and the probabilistic retrieval of similar events. Indeed, graphs are attached to statistical models for spatio-temporal processes, which at their turn describe physical changes in the observed scene. Therefore, we adjust a parametric model evaluating similarity types between graph patterns in order to represent user-specific semantics attached to spatio-temporal phenomena. The learning step is performed by the incremental definition of similarity types via user-provided spatio-temporal pattern examples attached to positive or/and negative semantics. From these examples, probabilities are inferred using a Bayesian network and a Dirichlet model. This enables to links user interest to a specific similarity model between graph patterns. According to the current state of learning, semantic posterior probabilities are updated for all possible graph patterns so that similar spatio-temporal phenomena can be recognized and retrieved from the image sequence. Few experiments performed on a multi-spectral SPOT image sequence illustrate the proposed spatio-temporal recognition method. Key-words: Pattern recognition; supervised learning, spatio-temporal phenomena, graph similarity; bayesian networks; Dirichlet model Apprentissage supervisé sur des graphes de similarité spatio-temporelle dans les séquences d’images satellites Résumé : Les séquences d’images satellites de haute résolution sont des signaux multidimensionnels composés de motifs spatio-temporels associés à des phénomènes nombreux et variés. Des méthodes bayésiennes ont été précédemment proposées dans [8] pour coder l’information contenue dans les séquences d’image satellitaire sous forme de graphes. Basé sur une telle représentation, ce papier présente une méthode d’apprentissage supervisé de sémantiques associées aux motifs spatio-temporels de ces séquences d’images. Cela permet la reconnaissance et la recherche probabiliste de phénomènes similaires. En effet, les graphes représentent des modèles statistiques de processus spatio-temporels, qui permettent de décrire des changements physiques observés dans la scène. En conséquence, par apprentissage supervisé, un modèle paramétrique évaluant les types de similarité entre motifs de graphes est ajusté pour représenter les sémantiques associées à ces phénomènes spatio-temporels. L’apprentissage est effectué par la définition incrémentale de types de similarités via des exemples fournis par l’utilsateur de motifs associés à des sémantiques positives ou/et négatives. A partir de ces exemples, des probabilités sont déduites par l’utilisation d’un réseau bayésien et d’un modèle de Dirichlet. Ces probabilits permettent de relier l’intérêt de l’utilisateur à un modèle de similarité spécifique entre motifs de graphe. A chaque stade d’apprentissage, les probabilités a posteriori sont actualisées pour l’ensemble des motifs de graphe possibles afin que les phénomènes spatio-temporels puissent être reconnus et retrouvés dans la séquence d’image. Quelques expériences efffectuées sur une séquence multi-spectral d’images SPOT illustrent la méthode de reconnaissance spatio-temporelle proposée. Mots-clés : Reconnaissance de forme, apprentissage supervisé, phénomènes spatio-temporels; similarité de graphes; réseaux bayésiens; modèle de Dirichlet Supervised learning on graphs of spatio-temporal similarity in satellite image sequences 3
منابع مشابه
معرفی شبکه های عصبی پیمانه ای عمیق با ساختار فضایی-زمانی دوگانه جهت بهبود بازشناسی گفتار پیوسته فارسی
In this article, growable deep modular neural networks for continuous speech recognition are introduced. These networks can be grown to implement the spatio-temporal information of the frame sequences at their input layer as well as their labels at the output layer at the same time. The trained neural network with such double spatio-temporal association structure can learn the phonetic sequence...
متن کاملBayesian Learning on Graphs for Reasoning on Image Time-series
Satellite image time-series (SITS) are multidimensional signals of high complexity. Their main characteristics are spatio-temporal patterns which describes the scene dynamics. The information contained in SITS was coded using Bayesian methods, resulting in a graph representation [2]. This paper further presents a concept of interactive learning for semantic labeling of spatio-temporal patterns ...
متن کاملMulti-Temporal Assessment of Mangrove Forests Change in the Coastal Areas of Bushehr Region Based on Landsat Satellite Imagery
Continual access to precise information about the land use/land cover (LULC) changes of the Earth’s surface is extremely important for any sustainable development program in which LULC serves as one of the major input criteria. In this study, a supervised classification was applied to three Landsat images collected in 1986, 1998and 2018, providing mangrove forests change data in the coastal are...
متن کاملRevealing the impact of changing land use of the annual spatiotemporal boundary layer height (Kermanshah Case Study)
Introduction Atmospheric boundary layer (ABL), is the lowest part of the atmosphere. Its behavior is directly influenced by its contact with earth surface. On earth it usually responds to changes in surface radiative forcing in an hour or less. In this layer physical quantities such as flow velocity, temperature, moisture, etc., display rapid fluctuations (turbulence) and vertical mixing is st...
متن کاملSpatio-temporal distribution of off-shore ships in the Pars Special Economic Energy Zone based on satellite imagery
Special Economic Zones (SEZs) are areas controlled by specific legislations so as toattain economic prosperity. These zones are commonly established and controlled bygovernment officials and are primarily characterized by growing population and developingtransport infrastructure. One relevant case is the Pars Special Economic Energy Zone(PSEEZ) situated in the south of Iran, on the northern sho...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- CoRR
دوره abs/0709.3013 شماره
صفحات -
تاریخ انتشار 2007