Advanced Materials Research Vols. 945-949

Abstract: This paper presents a content-oriented image metadata model. It aims to provide a standard metadata specification focusing on describing the information relevant to contents and topics of images, and enable different kinds of users to label an image in an effective and unified way. The model is also a key part of rich tag system (RTS), with which we can easily use controlled tags to fill metadata properties. RTS and the image metadata model have been applied to some practical applications, and been demonstrated to be effective and worthy of spreading.

Abstract: In this paper, a new yarn appearance measurement system based on machine vision is introduced. The yarn images are continuously captured by image acquisition system. To extract the main body of the yarn accurately, the yarn images are processed sequentially with threshold segmentation and morphological opening operation. Then the coefficient of variation (CV value) of diameter is calculated to characterize the yarn evenness. The measurement process achieves result (CV value) which can be compared with USTER evenness tester by image processing techniques. By comparing different methods which use different algorithms, a suitable method is chosen to be used in this new measurement system. Then a more accurate, more efficient and faster measurement system will meet requirements in the future.

Abstract: With the development of intelligent driving technology, recognition the vehicle in front of our cars became the hotspot in the field of intelligent driving research. This paper presents a self-adaptive front vehicle recognition algorithm with some unique improved method on the basis of analyzing and comparing the popular vehicle detection algorithm of domestic and foreign. Using the gray feature, vehicle shadow feature, taillights feature, license plate color domain feature and other features, the recognition algorithm can detect the vehicle in front of cars effectively, find out the safe passage area and avoid the potential risks. Finally, the feasibility of the algorithm is verified by experiment results with MATLAB tools.

Abstract: A new method is proposed to calculate the background in video sequences. The optical flow is estimated to determine the local regions occupied by moving objects. The background image is calculated by an efficient averaging process excluding the moving object regions, which overcomes the foreground-occluding problem in direct averaging method for background estimation. The experiments for traffic video processing prove the method’s effectiveness and robustness.

Abstract: The face detection has been a very important issue, the use of local and global face similarity between faces can be detected. In this paper, based on fractal image compression theory, we construct a local iterated function systems as a description of the face to detect the face.

Abstract: Real-time detecting information marked on billets is important for automatically manufacturing and management in steelworks. But due to the tough production environments in steel enterprises, capturing and identifying characters marked on hot billets have many challenges. This paper presents a real-time image capturing and segmenting method with machine vision for characters marked on hot billets, and characters area is located based on color information of images. Furthermore, considering the marked characters are often slant, we proposed a kind of characters skew correction method to adjust the alignment of characters, and then segment characters into singles for recognition. Finally, with the proposed method, we have conducted some experiments in Baosteel Company. The result shows that our method can achieve 97% segmentation rate if we select proper image acquisition device and preprocessing algorithm. Additionally, it provides a new way for steel enterprise real-time capturing and segmenting marked characters image.

Abstract: Pedestrian detection has a broad application prospect in automotive assisting driving system, but the real time performance is very poor in most common used detection methods. This paper presents a fast algorithm to realize the real-time pedestrian detection. The Local Binary Patterns (LBP) is used to describe the local texture information with the feature of less calculation, the HOG classifier to extract a typical feature of pedestrian’s edge, and then SVM to train and classify on the databases of INRIA and MIT. While scanning the images, interest regions are extracted to speed up the detection. Series of experiment results shows that the proposed pedestrian detecting strategy is effective and efficient.

Abstract: The saliva pregnancy test apparatus is dependent on manual operation to obtain results by observing the saliva crystallization image with complex operation, how to recognize the saliva crystallization image quickly and accurately has become an important research topic. In this paper, an image recognition method on the crystallization of saliva is proposed. Firstly, gray processing on the original image, improved Otsu method is used to select the threshold and binaryzation. Then the proportion of black and white pixels is calculated to identify saliva crystal images with the percentage values. The results show that the method is simple and treatment results are accurate.

Abstract: In crack image recognition, Donoho’s universal wavelet threshold de-noising method appears "over-kill" phenomenon due to the lack of self-adaptability of threshold value; hence the image may lose its edge details. To handle this problem, the Donoho’s universal threshold and threshold function are improved and an adaptive determination method of threshold coefficient is introduced in this paper. Experimental results shows that the proposed method can effectively remove digital image noise and achieve a better edge protection, higher edge preservation index, better visual effects and higher peak signal-to-noise ratio.

Abstract: Directionlet transform is a lattice-based skewed discrete wavelet transform. It has advantages of multi-directional and anisotropy compared with standard two-dimensional wavelet transform, thus, it is better at describing the characteristics of images. For the research focus of different-source image fusion, a novel fusion algorithm based on Directionlet transform was proposed, and the fusion speed was improved efficiently by combing the transform with a lifting scheme. Firstly, between transform direction and alignment direction, two registered source images were decomposed by using lifting Directionlet transform respectively in different times, thus anisotropic sub images were obtained. Then, the low frequency components were combined averagely and the selection principle of high frequency sub images were based on which has stronger anisotropic edge information. Finally, the fused image was obtained by using inverse Directionlet transform. Experimental results show that the fusion effect and speed are both better than standard wavelet transform and other second generation wavelet transform.