Advanced Materials Research Vols. 532-533

Abstract: In the arid areas of northwest China, sand-mulched field is an indigenous dry farming technique to reduce evaporation and conserve the limited rainfall for reliable crop production. The numerical simulation was carried out for analysis of the rainfall infiltration and evaporation process of sand-mulched field. Simulation results indicated that there were signifecant differences between the sand-mulched soil and that without mulch (control). Sand mulch on surface of soil may be effective in increasing infiltration and reducing runoff and evaporation. When the rainfall intensity is less than soil saturated hydraulic conductivity, sand mulch has little effect on the movement of soil infiltration wetting front. When the rainfall intensity is greater than the saturated permeability, the wetting front speed in sand-mulched soil is faster than that in bare soil during infiltration. During the simulation time, the runoff of sand-mulched soil is less than that of soil without mulch, variation in water potential of the sand-mulched soil is only 11% of that of the bare soil under the condition with evaporation intensity 0.6mm/h. Sand mulch can reduce water movement out of the soil by reducing the evaporation rate.

Abstract: Urban information is a kind of multi-sources data, the paper is focused on data indexes construction, classified indexes for non-spatial data, R-trees index for spatial data and the paper puts forward an area hiberarchy index tree structure to build up direct relationship between spatial data and non-spatial data for combined queries, and the experiment shows that the hiberarchy index tree is much validated and something useful is obtained.

Abstract: This paper introduces the design of a text categorization system based on Support Vector Machine (SVM). It analyzes the high dimensional characteristic of text data, the reason why SVM is suitable for text categorization. According to system data flow this system is constructed. This system consists of three subsystems which are text representation, classifier training and text classification. The core of this system is the classifier training, but text representation directly influences the currency of classifier and the performance of the system. Text feature vector space can be built by different kinds of feature selection and feature extraction methods. No research can indicate which one is the best method, so many feature selection and feature extraction methods are all developed in this system. For a specific classification task every feature selection method and every feature extraction method will be tested, and then a set of the best methods will be adopted.

Abstract: The greatest feature of cloud-based mass customization service platform is massive parallel processing of information. So the good or bad of the parallel processing determines the precision and efficiency of cloud-based mass customization service platform. This paper brings up the method of parallel processing of configuration design for the cloud-based mass customization service platform. The parallel processing of product configuration is performed by ABC analysis, parallel BOM and product family. Finally this research has been applied for customization product: Gearbox, and has improved the efficiency of parallel processing for the cloud-based mass customization service platform.

Abstract: The research of perception time is part of perception model, establish the perception based on depict, calculate the response time and duration of the process of perception, analysis the factor, point of direction of the future research.

Abstract: Preserving meaningful details such as blurred thin edges and low-contrast fine features is important in image de noising. A new method based on improved anisotropic diffusion model and wavelet transform is presented for image denoising. The proposed diffusion model incorporates both local gradient and gray-level variance to preserve edges and fine details while effectively removing noise in low-contrast surface images.

Abstract: In order to standardize the methods of information collection and sharing for nature reserves, a WebGIS based information system for nature reserve is designed and implemented in this paper. Large-scale information is administrated in this system, which can be conveniently accessed by information query, statistical analysis and data sharing.

Abstract: While 3DTV technology achieved great development in recent years, 3DTV program production can’t be wildly employed because 3D production is more complex than traditional 2D production. Stereoscopic camera parameters need to be well controlled, because the perceived depth range of stereoscopic display is limited by human factors and stereoscopic image should be kept within this limitation, so that the viewer can experience 3D video without any uncomfortable feeling. Recent stereoscopic image acquisition methods provide precise control to map depth range from scene to screen, but they don’t suit some shooting cases. In this paper, we present two new stereoscopic image creation methods, which are more practical than traditional ones. Additionally, our production methods are presented in two cases: (1) camera separation is unfixed; (2) camera separation is fixed. In the end, experiments are presented and the results show our methods can control the perceived depth very well.

Abstract: During video transmission over error prone network, compressed video bit-stream is sensitive to channel errors that may degrade the decoded pictures severely. In order to solve this problem, error concealment technique is a useful post-processing tool for recovering the lost information. In these methods, how to estimate the lost motion vector correctly is important for the quality of decoded picture. In order to recover the lost motion vector, an Decoder Motion Vector Estimation (DMVE) criterion was proposed and have well effect for recover the lost blocks. In this paper, we propose an improved error concealment method based on DMVE, which exploits the accurate motion vector by using redundant motion vector information. The experimental results with an H.264 codec show that our method improves both subjective and objective decoder reconstructed video quality, especially for sequences of drastic motion.

Abstract: This paper presents a paratactic spatial-temporal 2dimension data fusion model based on support vector machines (SVM) for traffic volume prediction of the abnormal state. Time and space SVM operates respectively in two parallel operating system models to reduce the time cost. By comparing the prediction results with which obtained by the multiple regression prediction method, the prediction accuracy is greatly improved by utilizing the paratactic spatial-temporal dimension data fusion model. Especially in the abnormal state caused by unexpected events (such as: traffic accidents, traffic jam etc), the proposed method can also significantly avoid structural system error of one-dimensional time source data fusion.