Earticle

As to solving the effectiveness and efficiency problem in the process of detecting the near duplicated video, we the thesis proposes a graph-based matching algorithm on sub-sequence of near duplicated video. The method will built similarity researching results based on the features of key frames feature into the query matching diagram, and then the near duplicated video detection is converted into a problem of searching the longest path in the matching results graph. As for its main advantage, firstly, it can find the best matching sequence in many cluttered matching results, which can effectively exclude a lot of noises brought by certain false “high similarity” matching, thus to some extent it can compensate the deficiencies of the underlying characterization force. Secondly, because it fully understands and uses the timing characteristics of the video sequences, the positioning accuracy of near duplicated video is with a high degree. Finally, multiple discrete paths existing in the matching results graphs are automatically detected, thus the situation where two video segments may exist several near duplicated videos can be detected once time. Experimental results show that the graph-based matching algorithm on video sub sequence improve the detection accuracy, at the same time improve the detection efficiency, which achieved good practice effect.

A digital signature is a mathematical structure for indicating the validity of digital information or any document. A message is created by a known sender whose digital signature provides a recipient reason, such that the sender cannot reject having sent the message confirmation and that the message was not changed in transportation integrity. The Signature recognition and verification are a behavioral biometric. It can be operated in two various types: one is the Off-Line or Static Signature Verification Technique and another is the On-line or Dynamic Signature Verification Technique. In this paper, we are studying about Off-Line or Static Signature Verification Technique. In this method, users write their own signature on the blank paper and then digitize it with an optical scanner or a camera, and then the biometric system identifies the signature by analyzing its shape and this collection is also called as “off-line” Signature verification. Signature authentication can be divided into three main classes. These classes are based on how alike a forgery is in relation to signature and are identified as random, simple and skilled. In the random forgery the forger does not know about the signer’s shape or signature name. In the simple forgery or unskillful forgery, the forger knows the name of the actual signer but don’t know how his signature looks like. And in the skilled forgery, the forger knows both the information of the signer.

UAV remote sensing, as a new method of remote sensing, has the characteristics of higher spatial resolution, fine timeliness and high flexibility. It is widely used in the field of natural disaster monitoring, urban planning, resource investigation, and has become one of the indispensable method of remote sensing data acquisition. However, because the UAV remote sensing platform is limited by the flight height and focal length of camera, the acquired image size is smaller, single image can’t cover the entire target area. Therefore, image mosaic has become a key technology to solve the problem. Image matching and image fusion are the key techniques of image mosaic. Due to the good robustness of image scaling, translation and rotation, this paper uses the SIFT algorithm to realize image matching of UAV. Since the feature extraction may produce false matches, RANSAC algorithm is applied to the feature point purification points. According to the seam-line in jointing overlap region, weighted fusion algorithm is applied to realize the image seamless splicing.

This paper proposes a novel approach to prepare an Artificial Sensory Head which can recognize senses (speech, vision) and change its facial expression according to it. These are basic human expressions like happy, sad, angry, fear, surprise and disgust.The ASH receives input from user which may be in the form of speech or image. On receiving input from user of the system the type of emotion is detected. A bit pattern corresponding to each emotion is set. This generated bit is sent to the target application through usb port. The microcontroller is programmed with the logic to activate the driver at the end. In our methodology, in the end a face model is implemented as driver. This face model has some optical device which is placed at the positions where change would occur during change in facial expressions. These locations are identified from pre determined motion vector of still images and optical devices are placed accordingly. On detection of emotion, an impulse is sent to these optical devices and the ones placed at the points which changes during change in facial expression would glow denoting the direction and location of change in emotions. All throughout the process, we have an face model that can be activated by speech and image.

This paper deals with the ambiguity problem of blind source separation (BSS) in the case where continuously received mixture signals are split in time and processed block by block. Due to the inherent permutation and scaling ambiguities of BSS, tying the separated components at each adjacent time blocks doesn’t recover the original source signals correctly in general. Inspired by the Permutation Method of reconstructing source signal blocks in time domain, a new ambiguity elimination approach is proposed in this paper. This method aims to concatenate the separated components in adjacent blocks by artificially setting contrast blocks for each adjacent time blocks. The core idea of this method is to utilize the associativity between components recovered from contrast blocks and corresponding adjacent blocks. Compared with Permutation Method, the main advantage of this new method consists in the fact that it is much more efficient in terms of separation quality and computational speed. Besides, a tradeoff can be adjusted between separation quality and computational speed by choosing different length of contrast blocks. Real-life experiments are performed to validate the performance of this method on the wireless communication system with two transmitting and receiving antennas.

This paper proposes an efficient multicast search scheme based on bipartite graph matching model, aiming at one of the most important problems in the multicast service, how can the wireless network track and locate the mobile uses in the idle state, under the tight bandwidth and delay constraints,. By quantifying the location uncertainty of mobile users with Shannon's entropy, the scheme adopts the LZ78 compression algorithm to design location update scheme and predict the location probability, in order to reduce the cost of location update. The multicast search system need to guarantee the maximum total probability that each user resides at the assigned paging area, for the purpose of an optimal performance on both expected paging delay and paging cost. Therefore, the bigraph-based multicast paging scheme (BMPS) firstly builds the bipartite graph model of multicast search problem, through converting the location probability into weight of the edge. BMPS decides the optimal allocation between the mobile users and location areas, which is mainly achieved by the maximum weight perfect matching in bipartite graph, while modifying the weights dynamically. Simulation results show that BMPS yields a low search delay as well as a low search cost, and reduces the impact of user collision.

In image processing, computer vision and pattern recognition, the Image retrieval is a most popular research area. In this paper, performance of various CBIR systems, based on combined feature i.e., color texture and shape, are compared.

The traditional information extraction methods based on specific domain usually depend on the domain dictionaries to discover the text feature. It is inconvenient for reproducing and difficult to transplant in multi-domain environment. The application scope is limited seriously. Oriented to the deficiencies above, a multi-domain web text feature extraction model for e-Science is proposed (named e-FTM). This model adopts the Chinese split words technology without dictionary into the process of multi-domain text feature discovery and avoids the dependency of domain dictionaries effectively. With the help of classification of common and individual features, the model tracks the generation and the development trend of domain events dynamically, and forms a couple of local data centers eventually. Through cooperative scheduling the domain knowledge between different local data centers, the knowledge utilization efficiency of the domain information in the global scope is improved sharply. To validate the performance, the experiments on the multi-domain text feature extraction, topic features dynamical tracking and the domain knowledge cooperative scheduling demonstrate that the model has higher application validity and practicality in e-Science environment.

With the rapid development of the Internet, the application of data mining in the Internet is becoming more and more extensive. However, the data source’s complex feature redundancy leads that data mining process becomes very inefficient and complex. So feature selection research is essential to make data mining more efficient and simple. In this paper, we propose a new way to measure the correlation degree of internal features of dataset which is a mutation of mutual information. Additionally we also introduce Hoeffding inequality as constraint of constructing algorithm. During the experiments, we use C4.5 classification algorithm as test algorithm and compare HSF with BIF(feature selection algorithm based on mutual information). Experiments results show that HSF performances better than BIF[1] in TP and FP rate, what’s more the feature subset obtained by HSF can significantly improve the TP, FP and memory usage of C4.5 classification algorithm.