Research on Hyperspectral Imagery Land Cover ClassificationMethod based on AdaBoost Improved Random Forest

In order to improve the classification accuracy of hyperspectral images,a new weighted random forest method based on AdaBoost is proposed.In this method,the concept of sample weight is introduced,and then the weight of each sample will be adjusted according to whether the sample is correctly classified.Those misclassified samples will be given higher weight value,to attract more attention of the classifier to improve the classification.Furthermore,the method gives the voting weight to every basic classifier according to their classification error rate.The basic classifier with higher classification accuracy will obtain larger voting weight.Two sets of Hyperspectral data(The CASI Hyperspectral Data acquired in Heihe region and CHRIS Hyperspectral Data acquired in the Yellow River Estuary) are used to verify the validity of the method.The results show that the weighted random forest has a better performance than the equal weight random forest and the SVM method in the overall classification accuracy,the average classification accuracy and the Kappa coefficient,which proves the efficiency of the proposed method.     

A Review of Lake Dynamic Change Researchbased on Remote Sensing

The shrink and expansion of lakes can reflect the regional changes in climate and environment. It has an important significance for further research of the climate change and the sustainable development.The rapid development of remote sensing technology has provided technical support for the dynamic change to real-time monitoring of lakes.This paper discussed the selection of data source,the delineation of lake water,the lake variation trends and causes.Then made a systematic summary of the current situation and progress in the studies on lake change and predicted the trends of lake change research in the future.     

Research on Transition Test from IPv4 to IPv6

Despite the phenomenal success of IPv4, it has certain limitations for wireless, QoS （Quality of Service） and streaming media. The development of IPv6-the next generation protocol to step in placo of the current Interact Protocol-has reached the level of being suitable for practical test and usage. This paper describes the test scenarios of the on-going transition from IPv4 to IPv6, which test bed network. We give some suggestions in IPv4/IPv6 transition at last.     

Analysis of Landscape Characteristics of Urban Heat /Sink Islandin Hangzhou based on ASTER Remote Sensing Imagery

Rapid urbanization has significant contributions to the Surface Urban Heat Island (SUHI).Analyzing the SUHI distribution and its impact factors using remote sensing data has received increasing attentions in the past decades,whereas few study has investigated that of the surface Urban Heat Sink Island (SUHI).The paper selects Hangzhou metropolis as a case study to explore SUHI/SUHS spatial patterns and its causes.We first retrieve the Land Surface Temperature (LST) using ASTER thermal infrared remote sensing imagery and extract the region of SUHI/SUHS using the Mean\|Standard deviation method.Landsat8 OLI data is used to classify land use and extract both impervious surface and vegetation information.After that,different landscape patterns within SUHI/SUHS area are analyzed and quantified by using several selected landscape index.The largest impact factors in SUHI/SUHS areas are identified.Finally,we analyze the spatial characteristics of LST using the spatial gradient analysis method,and reveal its relationship with vegetation and impervious surface.The results show that：(1) a large landscape pattern difference exists within SUHI/SUHS area;the impervious surface has the greatest impact on LST of the SUHI area,whereas the vegetation has more obviously cooling effect on LST of the SUHS area than the water body;(2) with the increasing distance from the city center,the same trend was found between the mean LST values and the impervious surface density (positive correlations),whereas the opposite trend between the mean LST values and the vegetation density (negative correlations).And the warming effect of impervious surface is greater than the cooling effect of vegetation in Hangzhou.     

Research on Enhancing the Functions of Animation in Web 3D

VRML offers powerful tool for generating 3D scene and realizing real time interactive animation on Web. This paper briefly introduces the related techniques of implementing interactive animation in Web 3D, focusing on how to use JavaScript to create Script node for extending VRML functions. This is very useful for virtual product design, product performance simulation, virtual assembly and remote education.     

Snow Cover Variation and Meteorological Factor Researchin Yarlung Zangbo Basin of Tibet from 2002 to 2015

The snow cover variation in Yarlung Zangbo from July 2002 to April 2015,as the mother river of Tibet,has been analyzed in this article.The hydrological year is from September of the year to August of the next year.The snow cover variation was studied used by cloudless snow cover data at 500 m and surface temperature from MODIS,digital elevation model (DEM) and meteorological data.The relationship was established between surface temperature from MODIS and major meteorological factors.The results show that the average snow cover in the study area is 3.56× 10⁴km² in the past 12 years,showing an overall decreasing trend.The area with large snow cover is in the upper and lower reaches of the basin,and the stable and seasonal snow dominate the rural area in the middle reaches and valley basins,with perennial less snow distribution.The stability of snow and perennial snow distribution is more than 4 000 meters above the sea level,which the snow lasts for more than 180 days.And the seasonal snow is mainly less than 4 000 meters above sea level with large variation of snow cover in the area.The inversely relationship is between the air temperature and the area of snow cover,but not significant between the precipitation and the area of snow cover.And the correlation in the surface temperature and the area of snow cover is obvious.In Short,the rising temperature is the main driving factor of snow area reduction in the basin.     

Research and Implementation for License Plate Recognition Based on Improved Projection Algorithm简

This paper analyzes and dissertates the discrete wavelet transform and improved projection algorithm in four kernel stages （image preprocessing, license plate localization, character segmentation, license plate recognition, i.e.） of license plate recognition system in detail. At last, it gives some conclusions and suggestions on future research.     

Extraction and Analysis of Hyperspectral Data and Characteristics from Pedicularis on Bayanbulak Grassland in Xinjiang

Through the field acquisition of three vegetation spectral datas,flowering Pedicularis,non flowering Pedicularis and common vegetation on Bayanbulak grassland,the first derivative,the two derivative and the reciprocal logarithm transformation were used to the smoothed and denoised data to analyze the difference sensitive bands of vegetation.The results showed that in visible light,non flowering Pedicularis and common vegetation showed the overall consistency,but the spectral curve of flowering Pedicularis showed a significant difference.In the red band and near infrared band at 750nm,non flowering Pedicularis reflectance increased significantly,and the three kinds of spectral reflectance showed significant differences.The reciprocal logarithmic transformation in the visible 580~680 nm band could be used to distinguish the Pedicularis as sensitive area.The spectral reflectance difference between the three at 655 nm was the most obvious.That solved the non flowering Pedicularis and common vegetation confusable problems.The improved normalized difference vegetation index by calculation,to further validate and showed the reciprocal logarithmic transformed values NDVI RLR could be distinguished the difference of flowering Pedicularis,non flowering Pedicularis and common vegetation.The extraction and analysis of hyperspectral data and characteristics from Pedicularis provided a theoretical basis for remote sensing monitoring of Pedicularis,and Remote sensing technology has great significance in Pedicularis resource survey and monitoring application.     

Error Sources and Post Processing Method for Land Use/cover Change Estimation of Dongguan Citybased on Landsat Remote Sensing Imagery with SVM

Urbanization is the world developing trend in the past century,which significantly changed the land use/cover of the urbanized area,and caused a series negative impacts,such as water shortage,flood increase,environment pollution,ecosystem degradation.How to estimate the land use/cover change more accurately has the prerequisite of studying the urbanization processes and its impacts,and is the research hot and challenge of the remote sensing and application communities.Dongguan city expressed the rapidest urbanization in China since China’s reform and opening door,and transferred from an agriculture county to a modern international metropolitan in less than 30 years,which has made a miracle in the world urbanization process.To prepare a high accuracy land use/cover change dataset for studying Dongguan’s urbanization process and its impacts,this paper first estimated the land use/cover change dataset by employing Support Vector Machine auto\|classification algorithm based on 12 Landsat remote sensing imageries from 1987 to 2015 at an average interval of 3 year.Then the error sources is analyzed by comparing the results estimated by using auto\|classification algorithm and visual interpretation,and a post data processing algorithm is proposed for refining the auto\|classification results.The final dataset of land use/cover change of Dongguan City is produced with the above method with an average accuracy of 86.87% and a Kappa coefficient of 0.83,which implies this product has a very good accuracy for analyzing the urbanization process of Dongguan city and its impacts.     

Fast Object Segmentation by Growing Minimal Paths from a Single Point on 2D or 3D Images

In this paper, we present a new method for segmenting closed contours and surfaces. Our work builds on a variant of the minimal path approach. First, an initial point on the desired contour is chosen by the user. Next, new keypoints are detected automatically using a front propagation approach. We assume that the desired object has a closed boundary. This a-priori knowledge on the topology is used to devise a relevant criterion for stopping the keypoint detection and front propagation. The final domain visited by the front will yield a band surrounding the object of interest. Linking pairs of neighboring keypoints with minimal paths allows us to extract a closed contour from a 2D image. This approach can also be used for finding an open curve giving extra information as stopping criteria. Detection of a variety of objects on real images is demonstrated. Using a similar idea, we can extract networks of minimal paths from a 3D image called Geodesic Meshing. The proposed method is applied to 3D data with promising results.