In recent years, Bayesian networks using unsupervised extracted image features have been applied in many remote sensing information mining systems to enable semantic-sensitive image retrieval. However, a simple Bayesian network insufficiently accounts for the spatial information, that is, the relations among image regions, for the semantic inference process. This drawback significantly impacts the retrieval performance, especially if the utilised features contain no or little spatial information. Therefore, this article proposes a context-sensitive Bayesian network, which infers semantic concepts of image regions based on the spectral and textural characteristics of the regions themselves as well as their contexts, that is, the adjacent regions. In order to compare the context-sensitive Bayesian network with the simple Bayesian network, comprehensive experiments were conducted based on high-resolution multispectral IKONOS imagery. The results show that the incorporation of the image regions' spatial relations not only significantly improves the accuracy of the semantic concepts inference, but also allows more flexibility in choosing the type of low-level features. 相似文献
During the last decade of the 20th century, extensive conversion in agricultural land use took place in Northeast China. The goal of this study is to ascertain its spatial distribution and regional differentiation, determine its causes, and analyze its environmental impact, Especially we attempt to elucidate how institutional constraints have facilitated the change at a time of agrarian restructuring when newly emerging free market was hybridized with the former planned economy. Information on six categories of land use was mapped from interpretation of Landsat TM images recorded in 1990,1995 and 2000. Most of land use changes took place during the first half of the decade, coinciding with abrupt and chaotic changes in government directives. Farmland was changed mainly to woodland,water body and built-up areas while woodland and grassland were converted chiefly to farmland.Spatially, the change from farmland to woodland was restricted to the west of the study area. The change from grassland to farmland took place in the grazing and farming interlocked west. These chaotic and occasionally conflicting changes were largely caused by lack of stability and consistency in agricultural land use policies promulgated. They have exerted adverse impacts on the local environment, including land degradation, increased flooding, and modified climate regime. 相似文献
Mining-induced tremors are indispensable events that gestate and trigger coal bursts. The radiated energy is usually considered a key index to assess coal burst risk of seismic events. This paper presents a model to assess coal burst risk of seismic events based on multiple seismic source parameters. By considering the distribution and relation laws of the seismic source parameters of coal bursts, the model aims to identify dangerous seismic events that more closely match the characteristics of multiple seismic source parameters of coal bursts. The new coal burst risk index T is proposed. It consists of the similarity index SI (representing the similarity degree of relations between seismic events and coal burst events based on seismic source parameters) and the strength index ST (representing the burst strength of seismic events). We studied 79 coal burst events that occurred during extraction in LW250105 of the Huating coal mine in Gansu Province, China. We obtained the distribution and relation laws of multiple seismic source parameters of coal burst events to establish SI and ST. Two groups of seismic events with different energy distributions were examined to compare the assessment results based on the new model and energy criteria. The results show that 80% and 89% of seismic events with strong coal burst risk in Groups A and B, respectively, were coincident, and the seismic events with medium coal burst risk were slightly less compared to those based on radiated energy. The results indicate that the assessment based on the T value is a modification and optimization of that based on radiated energy. This model is conducive to improving the efficiency of monitoring and early warning of coal burst risk.