Applied Mechanics and Materials Vols. 644-650

Abstract: It is very important that study the feature parameter extraction of bad point of wheat seeds based on image processing for judging the quality of wheat. Using image processing extract and analyze the collected images information, and based on the collected information analyze the bad point information of wheat seed, then extract the feature parameters. Traditional bad point’s feature extraction methods are completed by the manual operation, and the efficient is lower. Currently, by means of image processing technology can extract the bad point’s feature of wheat seed automatically. To this end, the research status of seed feature extraction based on image processing are reviewed and prospected. Experiments show that the method can better complete the bad point’s feature automatic extraction and recognition of wheat seeds.

Abstract: Utilizing visual characteristics of human eyes, color additive theory can make richer colors from red (R), green (G) and blue (B), the three primary colors. Our people live in a colorful world, and color exists everywhere. Many experiments has proved that the three colors can be mixed together to get various colors. In the paper, color additive theory on RGB is shown in a fun, novel manner by user-friendly dynamic process under Microsoft Visual C++ programming environment, whose purpose is to intuitively present the effect of different color overlapping to the viewers. Following the steps of reading image’s data, displaying the pixels, generating color bitmap and showing the dynamical movement of images in single channel, the program displays the mixing process of any two of red, green, and blue colors to get brighter colors, which are called yellow, cyan and magenta and an equal combination of the three colors can get white color.

Abstract: This paper presents our recent work towards the development of a public transportation query system based on speech error correction and natural language processing. The system enhances the accuracy of speech recognition by combining rules and word frequencies. This paper summarizes and classifies the error types of speech recognition in the field of transportation query, and then brings forth a semantic framework and a word-frequency-precedence mechanism with pinyin-text mapping. For the text after correctly recognizing, the system extracts the origin and the destination by forward-search-backward-match algorithm. And it synthesizes the least transferring algorithm and the shortest path algorithm to find a route. So it has characteristics of humanity and intelligence.

Abstract: Space-time two dimensional RAKE receiver was introduced into asynchronous and cooperative communication system, and a high-performance RAKE receiving algorithm was proposed. In this algorithm, the transmitting signals were estimated coarsely by traditional space-time rake receiver, then each multipath signal was reconstructed using the estimated result as well as the channel impulse response, and then eliminated successively from reception signals, so it can eliminate Inter Symbol Interference of receiver signals which generate by multipath fading, and the line of sight (LOS) component in the received signal was obtained, and finally, the space-time combination was carried out to the LOS component, and the spatial diversity was achieved. This algorithm can effectively reduce the bit error rate of the asynchronous cooperative communication system. Simulation result shows that the new algorithm can improve the performance of traditional space-time RAKE when SNR is higher than 5dB dramatically.

Abstract: In the process of image matching, it is involved such as image rotation, scale zooming, brightness change and other problems. In order to improve the precision of image matching, image matching algorithm based on SIFT feature point is proposed. First, the method of generating scale space is introduced. Then, the scale and position of feature points are determined through three dimension quadratic function and feature vectors are determined through gradient distribution characteristic of neighborhood pixels of feature points. Finally, feature matching is completed based on the Euclidean distance. The experiment result indicates that using SIFT feature point can achieve image matching effectively.

Abstract: An effective Human action recognition method based on the human skeletal information which is extracted by Kinect depth sensor is proposed in this paper. Skeleton’s 3D space coordinates and the angles between nodes of human related actions are collected as action characteristics through the research of human skeletal structure, node data and research on human actions. First, 3D information of human skeletons is acquired by Kinect depth sensors and the cosine of relevant nodes is calculated. Then human skeletal information within the time prior to current state is stored in real time. Finally, the relevant locations of the skeleton nodes and the variation of the cosine of skeletal joints within a certain time are analyzed to recognize the human motion. This algorithm has higher adaptability and practicability because of the complicated sample trainings and recognizing processes of traditional method is not taken up. The results of the experiment indicate that this method is with high recognition rate.

Abstract: This paper proposes a new method of applying nonbinary network coding in wireless broadcast network. This network consists of T(T>2) users who receive common information independently from a common base station (BS). We assume that the transmission blocks are subject to block-fading with independent fading coefficients for each block, nonbinary network codes over finite fields are used on top of channel codes.We take two kinds of schemes, with ACK and without ACK, into consideration. In the first situation while considering ACK mechanism, the forward error correction (FEC) and retransmitted techniques are used and network code (NC) can diminish the retransmit count significantly. In the second situation while utilizing broadcast nature without ACK, the NC could help to rebuild losing packets, thus decrease block error ratio.We analyze the delay performance and block error ratio, and verify the analytical performance with simulation. Both simulation and analysis confirm the advantage of proposed scheme.

Abstract: In order to improve the accuracy of recognition system for fatigue facial expression of driver, driver fatigue expression of this paper, the detection method for key feature points in the fatigue facial expression image of driver is applied in the paper to establish a fatigue expression image recognition model based on attention mechanism. Experimental results show that the algorithm can improve the recognition rate of driver's expression image, so as to record fatigue expression image of driver more accurate.

Abstract: Problem of identifying building crack is studied to provide technical support for the construction quality. In the building, building crackis an important factor affecting the construction quality. This paper presents a building crack recognition method based on cloud computing. Cloud model is established to pretreat the acquired building image, so as to improve image quality, and perform construction cracks identificationaccording to the processed image. Experimental results show that the improved algorithm for cracks construction identification can improve the recognition accuracy.

Abstract: This article introduces GPU-accelerated image processing parallel computing technology into standard core coding system of JPEG2000 static image compression and accelerates and designs the image compression process using CUDA acceleration principle. It also establishes the algorithm of image pixel array layered and reconstruction coding and realizes the coding of this algorithm using VC software. In order to verify the effectiveness and universal applicability of the algorithm and procedures, this paper compresses four images of different sizes and pixels in the static form of the JPEG2000. Through the comparison of the compression time, we can find that GPU hardware image processing system has a higher speedup ratio. With the increase of pixel and size, speedup ratio gradually increased which means that GPU acceleration has good adaptability.