Applied Mechanics and Materials Vols. 333-335

Abstract: A classical Lucas-Kanade optical flow algorithm was used to analysis the IR Image sequence of the wind-driven surface in this paper. Gaussian pyramid representation was introduced to retain both detail components and veracity for velocity field when considering the aperture problem and robustness. Three layers of pyramid for L-K optical flow is the best comparing with other layers (from one to four) in property. L-K optical flow algorithm mixed with pyramid representation shown an qualified power on calculating water surface flow field, demonstrated by optical flow fields on different wind speeds ( from 3m/s to 6m/s).

Abstract: The rough set is an effective tool to deal with incomplete, uncertain and fuzzy problem. In order to extract the edges information of the image by impulse noise pollution, further service for the follow-up image analysis and understanding, on the basis of the impulse noise pollution image preprocessing, an adaptive edge detection algorithm is put forward to process the impulse pollution image using rough set theory in this paper. Considering the target edge features can form a series of edge point constraints as the algorithm foundation,this method starts from the characteristics difference of the image edge and target area. Firstly, series edge constraint points are defined, and then the related questions of these conditions are solved by using rough set theory, so the edge detection algorithm is established. The experimental results show that this algorithm can effectively extract edge from the noise image information, to overcome the sensitivity of traditional algorithm for noise.

Abstract: Polynomial method in digital image geometric correction of a wide range of applications, this correction method does not consider the image spatial geometric process, but directly to the image itself is simulated, in order to achieve correction purposes, has simple and flexible features such as. This paper expounds the polynomial correction method, and combining with the features of MATLAB, designed and developed the digital image correction program. In order to verify the accuracy of the program to program, a series of image was corrected from experiment, result analysis, program design thinking is correct.

Abstract: Detection of preceding vehicles is a critical step in automotive driver assistance systems aiming to alert a driver about driving environments and possible collision with other vehicles. This paper developed an algorithm of detecting preceding vehicles based on computer vision, which exploits synthetically the vertical edge, shadow underneath vehicle, horizontal edge and symmetry features. The experimental results indicate that this algorithm is effective and precise.

Abstract: In this study a novel image processing approach is proposed to improve the denoising in SAR images based on wavelet packet and median filter, Median filter is adopted to remove noise in the wavelet packet domain. At first, the process of the novel is introduced in detail .At last, by adding some simulated noise in the SAR image, the performance of the proposed approach is shown and compared with other filter algorithms in terms of PSNR

Abstract: How to meet the customers requirement in product search is an import problem. Because the high-growing of e-commerce, a new demand emerges: the special-purpose search engine for searching goods from network shop. Our work focuses on the garment retrieval from the e-shopping database, which supports feature-based retrieval by shape categories and styles. This paper uses color and style characteristics of the garment images as a query information, based on the SIFT algorithm retrieval feature points of the garment image, and then using K-Means method to cluster the feature points. Use features to retrieve matching image from database of garment images.

Abstract: This paper introduces a micro-assembly system based on image processing and focuses on some key technologies. The system employed some orthogonal wavelet functions to filter images and the suitible filtering effects were obtained by Symlets wavelet basis. The auto-focusing method adopted in the micro-assembly system was the combination of coarse adjustment and fine adjustment. The resolution evaluation function for coarse adjustment was based on Krisch operator and the resolution evaluation function for fine one was based on high frequency component. Experimental results show that the focusing accuracy can reach ±4.0μm. A wavelet multiscale method based on variogram function was used in the edge detection, which can eliminate noise with more details retained. Generally the system can meet the requirements of micro-assembly.

Abstract: In order to restore the image in dust environment, a method was proposed to restore the images in dust environment based on a single still image. Firstly, an image degradation model considering multiple scattering factors caused by dust was built using the first-order multiple scattering methods. Then, a new geometric depth model of single image was presented. This model utilized depth map of a dust image to improve usual geometric depth model. According to the new model the every distance of the objects in dust environment was able to acquire from a single still image. Finally, an image in dust environment could be restored according to the relationship between the dark channel prior model and the distance information which was calculated by the geometric depth model. The experimental results have shown that the method enhanced luminance and contrast. The method provided a foundation for target recognition in the dust environments.

Abstract: Reconstruction of a 3D model for human knee joint is the basic step for its kinematics and dynamics analysis. To make further research on knee joint modeling, we present a new method to reconstruct 3D knee joint models based on magnetic resonance image (MRI). This method consists of steps as pretreatment of the images, the region growing for segmentation and the contour interpolation or the grey value interpolation and so on. The resulting 3D knee joint model are used for dynamics analysis of human knee joint after being imported into the finite-element platform which includes the tibia, the femur, the meniscus and the cartilages. The 3D model provides the possibility for the research on the movement roles and mechanics characteristics of the knee joint.

Abstract: Since the phenomena of intensity inhomogeneity in MR images are prominent and adversely affect quantitative image analysis .In this paper ,we propose a novel magnetic resonance (MR) image segmentation approach based on the kernel graph cuts technique .Because of automatic multiregion segmentation and global energy minimization ,the kernel graph cuts method can be applied to many kinds of images segmentation ,such as MR images and so on . To reduce or eliminate intensity inhomogeneity in MR images ,we add the intensity inhomogeneity correction step which is based on fuzzy c-means (FCM) algorithm before the segment procedure .Firstly , the real MR image data obtained after bias corrected by FCM algorithm .Secondly ,we segment the real MR image data by kernel graph cuts method .Experiments show that the kernel graph cuts method with intensity inhomogeneity correction have a better segment result in accuracy and over-segmentation .