Dongcheng Shi

Learn More
Blind image quality assessment refers to the problem of evaluating the visual quality of an image without any reference. In this paper, we present a no-reference blur metric for images and video using information contained in the image itself. We look at the sharpness of the sharpest edges in the blurred image, which contains information about the blurring,(More)
Because tongue image segmentation is an important procedure in the tongue characterization, its accuracy affects the following automatic process directly. In this paper, we propose a new tongue segmentation approach based on the combination a feature of tongue shape and the Snakes Correction model. We firstly get a rough tongue contour through using the(More)
In order to solve the pixel divergence problem brought by print and scan process, propose the algorithm that applying stretching contrast grade to adjust image pixel divergence. The parameter setting of this adjusting method uses peak signal noise ratio as standard of measurement, even though using different printer and scanner, we can also get good(More)
In this paper, an algorithm for moving objects segmentation based on MRF is provided. Firstly, initial segmentation for image sequences is introduced, and initial label field is gained. Then, according to the equivalence between GRF distribution and MRF distribution, corresponding energy function is formed. Finally, the label filed is obtained and the(More)
Markov random fields (MRF) can be used for a wide variety of vision problems. In this paper we will propose an algorithm of scenes feature description and analysis based on MRF. The theoretical framework is based on MRF and Bayesian estimation via the energy optimization. We analyze the texture feature. Using MRF to modeling on the image, then combine with(More)
An algorithm of moving object detection based on scene understanding is proposed. The knowledge of color is extracted by SOFM from images which contain various colors information, and is used to segment the detected scene images. The move information of original object and new object in the scene would be detected by scene understanding using the scene(More)
Donoho and Candes proposed the theory of compressive sensing which mainly based on the linear model, the core assumption is sparse signal. This theory has been successfully applying to process audio signal. It appears rapidly in the field of image processing (including face recognition), which raised the research upsurge of technology. Nowadays, the most(More)
A method of stereo matching based on binocular vision system is proposed in this paper. First, the paper adopts RAC camera calibration method rectified input images. Then, in the stereo matching section, we adopt Segment-Based stereo matching algorithm. There are four major steps of the proposed algorithm, namely, color segmentation, local matching in pixel(More)
The study and implementation of a binocular vision based on human face description. In the traditional face recognition, we based on the principle of human visual design of the double cameras to extract three-dimensional information. For the two cameras input video frame, according to correlative coefficient and polar constraints to find the corresponding(More)
  • 1