Accepted Papers

  • Sharp or Blur: A Fast No-Reference Quality Metric for Realistic Photos
    Fan Zhang,Lenovo Research, Shenzhen, China
    There is an increasing demand on identifying the sharp and the blur photos from a burst of series or a mass of collection. Subjective assessment on image blurriness takes account of not only pixel variation but also the region of interest and the scene type. It makes measuring image sharpness in line with visual perception very challenging. In this paper, we devise a no-reference image sharpness metric, which combines a set of gradient-based features adept in estimating Gaussian blur, out-of-focus blur and motion blur respectively. We propose a dataset-adaptive logistic regression to build the metric upon multiple datasets, where over half of the samples are realistic blurry photos. Cross validation confirms that our metric outperforms the-state-of-the-art methods on the datasets with a total of 1577 images. Moreover, our metric is very fast, suitable for parallelization, and has the potential of running on mobile or embedded devices.
  • Multi-Parameter Fusion Feature Extraction Method In Automaton Fault Diagnosis
    Pan Hongxia, Li Sha and Xu Xin,North University of China, China
    In order to comprehensively measure signal fractal characteristics, the override method is used to calculate the vibration signal generalized fractal dimension and draw the generalized fractal dimension spectrum, box dimension, information dimension, correlation dimension automaton as fault feature values are extracted. Then quantitative diagnosis index at the level of feature information integration — the index distance of three demensional characteristic parameters is proposed. In view of the fault feature parameters extracted under various conditions, We compute the average respectively, then obtain the four standard centers separately representing automaton four conditions in three dimensional space.In view of the vibration signals to be detected, according to the extracting three-dimensional characteristic parameters, we can find the corresponding characteristic index points in the threedimensional space, respectively calculate these distances of between the characteristic index points and four standard centers, the index distance of three demensional characteristic parameters, and draw graphs of the index distance of three demensional characteristic parameters to identify fault conditions intuitively. Singular spectrum entropy, power spectrum entropy, local wave spatial spectral entropy are extracted as quantitative features to describe the state changes of signal in time domain, frequency domain, time-frequency domain. Calculating the index distance of six-demensional characteristic parameters is suggested to identify conditions. Two graphs of the index distance of six-dimensional and three-dimensional characteristic parameters are drawed simultaneously to increase the comparative.Diagnosis results indicate that, the index distance of six-dimensional characteristic parameters can accuratly identify fault conditions of automaton, compared to the index distance of three-dimensional characteristic parameters.
  • Sampling Density Criterion for Circular Structured Light 3D Imaging
    Deokwoo Lee1 and Hamid Krim2, 1Youngsan University, South Korea, 2North Carolina State University, USA
    In general, 3D reconstruction work has chiefly focused on the accuracy of the reconstruction results in computer vision, and efficient 3D functional camera system has been of interest in the field of mobile camera as well. The optimal sampling density, or also referred to as the minimum sampling rate for 3D or highdimensional signal reconstruction is proposed in this paper. There have been many research activities to develop an adaptive sampling theorem beyond the Shannon-Nyquist Sampling Theorem in the areas of signal processing, but the sampling theorem for 3D imaging or reconstruction is an open challenging topic and crucial part of our contribution in this paper. We hence propose an approach to sampling rate (lower / upper bound)determination to recover 3D objects (surfaces) represented by a set of circular light patterns, and the criterion for a sampling rate is formulated using geometric characteristics of a the light patterns overlaid on the surface.The proposed method is in a sense a foundation for a sampling theorem applied to 3D image processing, by establishing a relationship between frequency components and geometric information of a surface.
  • A Flower Image Retrieval Method Based On Memetic Feature Selection Algorithm
    Zhijiao Xiao, Meiyuan Cao, Shaoyang Zhou, Zhiwei Liang, and Jianmin Jiang, Shenzhen University, China
    Content-based image retrieval is a hot topic, while contend-based flower image retrieval is one of the most important and challenging problem. In this paper, a flower image retrieval method is proposed based on feature selection using memetic algorithm. The memetic algorithm, which combines a global search strategy with a local search strategy, is used to select the optimal feature subset. Genetic algorithm is used as the global search strategy while approximate Markov blanket is adopted as the local search strategy. Primary classifiers are trained using the proposed memetic algorithm for each kind of features. The probabilities obtained by all the primary classifiers are combined together to form a mid-level feature used to train the final classifier. Experimental results show that the proposed method selects fewer number of features with better precisions and recall ratios. That also brings improvements on retrieval time.
  • A Binary to Residue Conversion Using New Proposed Non-Coprime Moduli Set
    Mansour Bader1, Andraws Swidan2 , Mazin Al-hadidi3 and Baha Rababah4, 1,2Jordan University,Jordan, 3Al-Balqa'a Applied University, Jordan, 4University of Portsmouth, UK
    Residue Number System is generally supposed to use co-prime moduli set. Non-coprime moduli sets are a field in RNS which is little studied. That's why this work was devoted to them. The resources that discuss non-coprime in RNS are very limited. For the previous reasons, this paper analyses the RNS conversion using suggested non-coprime moduli set. This paper suggests a new non-coprime moduli set and investigates its performance. The suggested new moduli set has the general representation as {2n–2, 2n, 2n+2}, where n ? {2,3,…..,8}. The calculations among the moduli are done with this n value. These moduli are 2 spaces apart on the numbers line from each other. This range helps in the algorithm’s calculations as to be shown.The proposed non-coprime moduli set is investigated. Conversion algorithm from Binary to Residue is developed. Correctness of the algorithm was obtained through simulation program. Conversion algorithm is implemented.
  • Non-local means image denoising with bilateral structure tensor
    Li Hua and Xu Yi, Wuhan university of technology, China
    Non-local means, bilateral structure tensor, image denoising, texture. Abstract. Non-local means image denoising with bilateral structure tensor algorithm is put forward for the reason that Non-local means(NLM) algorithm has a weaker detail retention and noise immunity. Different from the initial structure tensor, we can get better texture description between similar blocks by using of bilateral structure tensor of noise resistance and texture features. New texture can improve the description of NLM value calculation function which filter the noise of images. Compared with traditional NLM algorithm, NLM-BST algorithm gets better image detail preservation in noise immunity. The experimental results show that operator is validated in denoising and image detail reservations.
Copyright ® SPM 2016