Applied Mechanics and Materials Vols. 325-326

Abstract: Taking the compression issue of remote sensing images as the study subject, this paper analyses the technical process for JPEG2000 compression and the image features of remote sensing images, natural images and figural images, and puts forwards an integrative optimization algorithm for effective compression display for remote sensing images based on common JPEG2000 compression frame. Thereinto, it includes high-frequency component filtering treatment, parallel processing of bit plane coding pass scanning and improved ROI coding algorithm, which give obvious clews to both overall consumed time for image compression and the gradual display effect in ROI region.

Abstract: This paper presents a more accurate and faster method for sub-pixel TDICCD image registration. The method makes the best of the overlapping pixels between multi-channel images, which allows for scaling, translation and rotation. Secondly, the proposed technique combines an efficient pixel-moving interpolation scheme with surface fitting, which makes use of accurate interpolation calculation and fast surface fitting in the iterative process. Finally, the accuracy and speed of the algorithm is evaluated by sub-pixel registration of multi-channel images and comparison with other sorts of efficient methods. The experiment results show that the accuracy of the method reaches 0.01 pixels and it is 3 times faster than the interpolation method.

Abstract: Captured circular marks are deformed sometimes when Automatic Optical Inspection (AOI) is used to detect various defects on Printed Circuit Boards (PCB), which may affect the precision of inspection. A new accurate positioning method of circular marks is proposed to solve the problem by obtaining the center of the most round ellipse based on the criterion that the ratio of the difference between the length and width of its circumscribed rectangle and the width of the rectangle is less than 0.1. The simulation tests show that, if the mark has much more deformations, the center positioning error of the proposed algorithm is about 0.013 pixels, and the running time is less than 40ms. Therefore, the proposed method provides good characteristics such as speediness, strong anti-interference ability and robustness.

Abstract: This paper covers brain tumor diagnosis system based on image analysis and mining and its application. The system use the algorithm of fuzzy region competition, extracts shape feature factors to classify tumor shape, maps shape classification extracted automatically and other medical image features to numbers and then feed to bayesian network to sort the brain tumor automatically.

Abstract: Through the research of the association rules mining technology and Apriori algorithm, the defects are found in Apriori algorithm. In view of the deficiencies, an improved algorithm is proposed. The algorithm scans database only once, and efficiently reduces the I/O time. The matrix of frequent itemsets is used to store and reduce the transaction data, which saves the storage space. By comparison of Apriori algorithm and improved algorithm, the results of experiments show that the efficiency of the improved algorithm is increased. Finally, an application example of the association rules is given. The improved algorithm is introduced to book lending deal. The rules among the book-borrowed are discovered and analyzed.

Abstract: Be aimed at the problems that K-medoids algorithm is easy to fall into the local optimal value and basic particle swarm algorithm is easy to fall into the premature convergence, this paper joins the Simulated Annealing (SA) thought and proposes a novel K-medoids clustering algorithm based on Particle swarm optimization algorithm with simulated annealing. The new algorithm combines the quick optimization ability of particle swarm optimization algorithm and the probability of jumping property with SA, and maintains the characteristics that particle swarm algorithm is easy to realize, and improves the ability of the algorithm from local extreme value point. The experimental results show that the algorithm enhances the convergence speed and accuracy of the algorithm, and the clustering effect is better than the original k-medoids algorithm.

Abstract: As a relatively novel clustering approach, Particle Swarm Optimization (PSO) prevents k-means algorithm from falling into local optimum effectively, and has made relatively notable successes in clustering, however, using Hard C-Means algorithm when randomly obtaining initial clustering centers is required in most existing PSOs, while no definite limit existing in these samples actually. Based on this, we utilized an improved PSO; along with effective processing methods on boundary objects of Rough Set Theory, we proposed a new rough clustering algorithm based on PSO. It can adjust the upper and lower approximations weighting factors dynamically, and coordinate the proportions of upper and lower approximations in different generations as well. Finally, we compared it with several common clustering methods using Iris dataset of UCI. It turned out that the algorithm has higher accuracy and stability, along with better comprehensive performance.

Abstract: Images matching is the basis of image registration. For their difference, a improved SURF(speeded up robust features) algorithm was proposed for the infrared and visible images matching. Firstly, edges were extracted from the images to improve the similarity of infrared and visible images. Then SURF algorithm was used to detect interest points, and the dimension of the point descriptor was 64. Finally, found the matching points by Euclidean distance. Experimental results show that some invalid data points were eliminated.

Abstract: In order to improve the quality of image de-noising,against the disadvantages of the distortion caused by the method of the hard threshold de-noising and the fuzzy phenomenon of the details caused by the method of the soft threshold de-noising, this article proposes a new method of wavelet threshold de-noising for the ultrasonic images. It is indicated in the simulation result that this method has good de-noising function: it can remove the noise effectively and retain the details of the images and the edge information at the same time.

Abstract: This paper focuses on the constant modulus Busgang blind equalization algorithm (CMA blind equalization algorithm in Constant, The Modulus Algorithm). Analysis of the convergence performance of the traditional CMA blind equalization algorithm, the fixed step size, convergence speed and convergence of mutual constraint between the precision of its application under great restrictions is demonstrated in the paper. In order to solve this contradiction, this paper presents a CMA blind equalization algorithm based on the mean square error (MSE Mean Square Error).