In order to make full use of the spatial information of images in the classification of natural scene, we use the spatial partition model. But mechanically space division caused the abuse of spatial information. So spatial partition model must be properly improved to make the different categories of images were more diversity, so that the classification performance is improved. In addition, to further improve the performance, we use FAN-SIFT as local image features. Experiments made on 8 scenes image dataset and Caltech101 dataset show that the improved model can obtain better classification performance.
This paper proposed a bi-level programming model to optimize the locations and capacity for rail-based park-and-ride sites to promote transit patronage. A multinomial logit model was incorporated in a mode split/traffic assignment model to assess any given park-and-ride scheme. This model was then taken as the lower level model, and the upper level programming model is established to optimize the location and capacity of park-and-ride with the goal of promoting transit patronage. A heuristic tabu search algorithm is then adopted to solve this model.
With the development of high-voltage switches and high-speed RF circuits, the enhancement mode(E-mode) AlGaN/GaN HEMTs have become a hot topic in those fields. The E-mode GaN-based HEMTs have channel current at the positive gate voltage, greatly expanding the device in low power digital circuit applications. The main methods to realize E-mode AlGaN/GaN HEMT power devices are p-GaN gate technology, recessed gate structure, fluoride ion implantation technology and Cascode structure (Cascode). In this paper, the advantage and main realizable methods of E-mode AlGaN/GaN HEMT are briefly described. The research status and problems of E-mode AlGaN/GaN HEMT devices fabricated by p-GaN gate technology are summarized. The advances of p-GaN gate technology, and focuses on how these research results can improve the power characteristics and reliability of E-mode AlGaN/GaN HEMT by optimizing device structure and improving process technology, are discussed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.