Extraction of the eye features, like the pupil center and radius, eye comers and eyelid contours from the frontal face images often provides us useful cues in many important applications, such as face recognition, fac...
详细信息
The process of color separation of tongue image in traditional Chinese medicine (TCM) is decomposed into two steps: region partition and color classification. In the first step, a partition method based on JSEG is pro...
详细信息
To overcome the drawbacks of print measurements, a novel measurement approach, namely dot structure based measurement, is proposed. In addition, the structure feature extraction of microscopic dot is discussed. The pr...
详细信息
ISBN:
(纸本)0780382730
To overcome the drawbacks of print measurements, a novel measurement approach, namely dot structure based measurement, is proposed. In addition, the structure feature extraction of microscopic dot is discussed. The proposed scheme provides a precondition for such further research as print quality surveillance, fault diagnosis, color design and printing theories as well.
In this paper, we investigate how camera pose can be estimated from 2D to 3D corner correspondence when vertex of the corner is occluded. We show that the image coordinate of the occluded vertex can be easily estimate...
详细信息
In this paper, we investigate how camera pose can be estimated from 2D to 3D corner correspondence when vertex of the corner is occluded. We show that the image coordinate of the occluded vertex can be easily estimated from the image edge lines by an analytical method. Then the camera pose parameters can be recovered from 2D to 3D correspondences of the corner and a nonsingular point on it over a single view. A ground truth comparison is made by experiments using controlled scenes, which shows that the method is promising.
We propose a new efficient method to detect ellipses in binary or gray-scale images, called restricted randomized Hough transform (RRHT). The key of RRHT is to restrict the scope of selected points when detecting elli...
详细信息
We propose a new efficient method to detect ellipses in binary or gray-scale images, called restricted randomized Hough transform (RRHT). The key of RRHT is to restrict the scope of selected points when detecting ellipses by prior imageprocessing, from which the information of curves can be obtained and further work focuses on detection and validation of each curve orderly. Artificial images and natural gray images are employed to test this method and the results show that RRHT spends less time to detect ellipses more successfully compared with that of conventional randomized Hough transform (RHT).
In this paper, we present a new method for the detection of principal curves in complicated feature images. Based on the criteria of the shortest path of curves and directional deviation of paths, principal curve dete...
详细信息
ISBN:
(纸本)0769521282
In this paper, we present a new method for the detection of principal curves in complicated feature images. Based on the criteria of the shortest path of curves and directional deviation of paths, principal curve detection is carried out in graph domain. DFS searching scheme is adopted in exploration of a graph network. The motivation of this research is to find road boundaries and house contours from printed map images. By extensive experiments, the algorithm has shown good efficiency and robustness with real map images. The technique described in this paper can also be used in other applications, such as in character recognition, to separate characters from other unwanted document components that overlap with the characters.
We present a new method for the determination of camera pose from 2D to 3D corner correspondence. Two cases are considered: the orthogonal corner and the general corner with known space angles. The contribution of the...
详细信息
ISBN:
(纸本)0769521088
We present a new method for the determination of camera pose from 2D to 3D corner correspondence. Two cases are considered: the orthogonal corner and the general corner with known space angles. The contribution of the paper is in two folds: one is that the camera pose parameters, i.e., the rotation and translation, are easily recovered from a 2D to 3D corner correspondence; the other is that experiments using both simulated data and real images are conducted, which present good results.
We present a novel watermarking approach based on classification for authentication, in which a watermark is embedded into the host image. When the marked image is modified, the extracted watermark is also different t...
详细信息
We present a novel watermarking approach based on classification for authentication, in which a watermark is embedded into the host image. When the marked image is modified, the extracted watermark is also different to the original watermark, and different kinds of modification lead to different extracted watermarks. In this paper, different kinds of modification are considered as classes, and we used classification algorithm to recognize the modifications with high probability. Simulation results show that the proposed method is potential and effective.
Text that appears in a scene or is graphically added to video can provide an important supplemental source of index information as well as clues for decoding the video's structure and for classification. In this w...
详细信息
Text that appears in a scene or is graphically added to video can provide an important supplemental source of index information as well as clues for decoding the video's structure and for classification. In this work, a novel algorithm is present for detecting and locating text in digital video. The first module of the system divides an image into small blocks which are featured by pixel value and are fed to SVM (support vector machine) to classify' text blocks or not. The other module is to do post-processing on the classified text blocks to identify the rectangle region of them and OCR can be used further easily. Experiments conducted with a variety of video sources showed that our method could detect and locate text region successfully by SVM with comparatively less samples.
Extraction of video object plane (VOP) is the central challenge for content-based video processing. It enables separate and flexible manipulation, blending and arrangement of content at MPEG-4 based decoder. We presen...
详细信息
ISBN:
(纸本)0780384032
Extraction of video object plane (VOP) is the central challenge for content-based video processing. It enables separate and flexible manipulation, blending and arrangement of content at MPEG-4 based decoder. We present a fast and efficient extraction algorithm depending on edge detection. Our algorithm begins with finding edge candidates detected by Canny edge detector. After selecting edge pixels which belong to the moving object in current frame, the VOP is extracted by using morphological operations. The proposed algorithm is evaluated on MPEG-4 test sequence and generates desirable VOPs at a lower computation cost.
暂无评论