摘要:In visual surroundings, the mapping from hierarchical and multiscalar relationships into visual perception is showed as the visual propensities in analyzing and organizing the surroundings. The simplest relationships are local position and orientation relationships, and their corresponding visual propensities are the visual position and orientation orders. At phenomenological level, position and orientation orders are explicitly expressed in an energy model. Several K kinds of perceptual binding emerge as global solutions of this energy model. A general theory for perceptual organization including visual position and orientation orders, can be used to consider the other clues for perceptual organization, such as closeness, similarity, symmetry and so on. This theory provides a general approach to visual representation and learning.
摘要:In the applications such as remote sensing and computer vision, to reconstruct a high-resolution image from multiframes of low-resolution images is required. Former work was based o n the same r eso lution fr ames. T his paper is based on the theorem of spectral aliasing resulting from undersampled low-resolution image, and a general formula of the spectral aliasing
is proposed for the problem of different resolutions. A row-action method even for noisy and blurred frames are given in the DFT domain as well. Computer simulations demonstrate the performance of the algorithm.
摘要:A color saturation adjustment algorithm of video coding in MPEG-2 is given in this paper. This algorithm divides the color difference signals of current frame into four color saturation regions of very high, high, low, very low region by using neural networks and derive the modification factor to adjust the color saturation of the current frame. Simulation results
indicate that this method can improve the subjective and objectiv equality of video decoding frame in terms of decoding peak signal-to-noise ratio and color saturation.
摘要:Panoramic Video which uses 360 degree panoramic imaegs is a new approach for composing virtual environment. The panoramic images can be created by “stitching”together overlapping images taken with an ordinary camera. In this paper we put forward a valid stitching algorithm which utilizes the correspondent pixels' similarity of over lapping image portion to make the resulted images seamless and smooth.
摘要:This paper discusses the principle of image morphing, prestnts an algorithm, in which two-dimension image in square region is mapped to any two-dimension out line region, and gives some results of the algorithm by use of BMP file. Examples demonstrate that this algorithm is extremly valid.
摘要:Facial feature detection is crucial for visual recognition of human faces. This paper reports the result on a study in detecting face contours from front-view ID-type pictures. A simple and effective way is proposed to detect the eye and mouth positions. Based on the relevant subimage defined in Ref.5, the location and curvature of the chin line, together with the cheek lines can be detected and the shape of the face was characterized. Especially in this paper, the eye and mouth contour sex tractions become easier and more accurate.
关键词:Facial feature location and extraction;Hough transformation;energy function
摘要:In this article, an efficient video encoding approach is proposed to reduce the blocking effect. It employs a new logical realization structure of coefficient-by-coefficient adaptive DCT coding algorithm. Meanwhile, a simple but efficient adaptive controlling strategy is proposed to replace the exactly adaptation which is very complex. Also, to overcome the shortcoming that the performance of the adaptation decreases at low bit rates, an excellent fast algorithm for motion estimation which can efficiently improve the adaptive ability is proposed. The experimental results indicate that the performance of the new encoding approach has a marked improvement over the conventional DCT-based code.
摘要:Image compression could be regarded as a compromise between the bit ratio and perceptual distortion. Image quality evaluation plays a very important role in the design, optimization and evaluation of an image compression algorithm. Classical image quality evaluation methods such as MOS, MSE and PSNR can not represent distortion perceived by human eyes. New measurements are based on HVS (Human Visual System). They are generally defined as a function of Amplitude Nonlinear it y, CSF, Detection M echanisms. We build a general evaluation model on HVS by B spline wavelet and error signal segmentation technology. The future of image quality evaluation will be directed to psychology and physiology experiments, the simplification of HVS and generalization of evaluation model.
摘要:The anisotropic features (orientation) of texture is very important for texture analysis. In this paper we introduced a new multi-channel model for texture analysis and synthesis based on the idea of isotropic and anisotropy power decomposition. We give the algorithm to determine the dominant orientation of the texture and to approximate the remain image with stochastic model. The experiment results for texture synthesis with this model and the performance comparison with her approach show the efficiency and flexibility of this method.
摘要:Various major factors causing geometric distortions of airborne multispectral scanning images were analyzed, and a method to correct geometric distortions of airborne multispectral scanning images by using the orientation, movement and positional parameters of airplane recorded by Inertial Nevigation System is proposed. Test result indicates that compared with the traditional method-GCP, this method not only yields better accuracy, but also can be performed automatically by the computer.
摘要:This paper presents a practical method of Chinese fonts composition which uses uniform cubic B-spline interpolating curves to express Chinese stroke out line. It builds correspondence relationship between composited curves based on continuity and uniqueness of B-spline interpolating curve. If compositin ratio coeff. and unified control point's length of B-spline
curves and range of their parametic values are given, new fonts are generated by new stroke out line expression. Simulation results show it is a simple, easy and effentive method of font composition.
摘要:This is the third in the survey series of yearly bibliographies on image engineering in China. Considering the current status of the researches on image engineering and the applications of image techniques as well as the distribution of related publications in China, near 300 research and technique references are selected from more than 1400 papers published in a set of 15 Chinese journals, in which papers concerning image engineering are relatively concentrated. These references are classified into different categories according to their contents. An analysis and a discussion of the statistics made on the classification results are also presented. This work shows a general and up-to-date picture of the progresses of image engineering in China, supplies a convenient means of literature searching for readers doing researches and/or applications in this field, and provides a useful reference for the editors of journals and potential authors of papers.