International Journal of Image and Graphics

Image processing and analysis in fuzzy set theoretic framework is addressed. Various uncertainties involved in these problems and the relevance of fuzzy set theory in handling them are explained. Different image ambiguity measures based on fuzzy entropy and fuzzy geometry of image subsets are mentioned. A discussion is made on the flexibility in choosing membership functions. Illustrations of commonly used fuzzy image processing operations such as enhancement, edge detection segmentation, skeleton extraction, feature extraction are then provided, along with their significance and characteristics. Their applications to some real life problems, e.g., motion frame analysis, remotely sensed image analysis, modeling face images are finally described. An extensive bibliography is also provided.

Human face image processing techniques have many applications, such as in security operations, entertainment, medical imaging and telecommunications. In this paper, we provide an overview of existing computer algorithms for face detection and facial feature location, face recognition, image compression and animation. We also discuss limitations of current methods and research work needed in the future.

The main result of the paper is a new and efficient algorithm to compute the closest possible representable intersection point between two lines in the plane. The coordinates of the points that define the lines are given as single precision floating-point numbers. The novelty of the algorithm is the method for deriving the best possible representable floating point numbers: instead of solving the equations to compute the line intersection coordinates exactly, which is a computationally expensive procedure, an iterative binary search procedure is applied. When the required precision is achieved, the algorithm stops. Only exact comparison tests are needed. Interval arithmetic is applied to further speed up the process. Experimental results demonstrate that the proposed algorithm is on the average ten times faster than an implementation of the line intersection computation subroutine using the CORE library exact arithmetic.

This paper presents a stereoscopic video compression scheme using a novel fast affine motion estimation technique. A temporal scalable approach is used to achieve backward compatibility with a standard definition TV. We use an adaptive mode selection scheme from three temporal locations in both channels. Both block-based and affine-motion based approaches are used to achieve two levels of improvements with different complexities. An innovative motion estimation technique using Gauss–Newton optimization and pyramid processing is implemented to efficiently estimate affine parameters. Unlike other Gauss–Newton approaches, our search technique uses only addition, subtraction and multiplication and it converges within four iterations, which implies great complexity reduction. An efficient and robust affine motion prediction yields significant over the disparity-based approach. Part of the disparity-based approach has been tested in the rigorous MPEG-2 bitstream exchange process, and adopted in the MPEG-2 Multi-View Profile (MVP).

This paper presents an efficient and simple quad edge conversion method of polygonal (manifold) objects. In a wide variety of applications such as scientific visualization, virtual reality and computer aided geometric design, polygonal objects are expected to be visualized and manipulated within a given time constraint. To achieve these expectations, it is necessary to introduce an efficient data structure as well as high performance graphics hardware and real-time processing techniques such as simplification and level of details. The quad edge data structure is very efficient for handling polygonal objects even though it was originally designed to handle the subdivisions of manifold objects such as Delaunay triangulations and Voronoi diagrams. It, however, has not been used widely because there is no efficient algorithm for quad edge conversion of conventional polygonal objects.

In this paper, we propose a new incremental quad edge conversion algorithm that processes the triangles one by one. Since quad edge has only the splice as a topological operator, the quad edge conversion of each triangle is done by applying three splice operations, a splice per vertex. As an applicaion for the quad edge, a simplification of conventional polygonal objects is implemented. It includes the removing, moving, replacing, and inserting of vertices and edges.

Vessel tracking is usually needed for various medical usages such as clinical diagnosis and surgery planning. A clear understanding of the patient's actual vessel morphology is very important in planning the surgery. Tracking the vessels/bronchia from scanned medical images such as CT or MRI manually is very tedious and labor-intensive. On the other hand, a fully automatic tracking algorithm may not be very reliable. It may be sensitive to the noise in the volume data. In this paper, we propose a vessel tracking method that finds air-way paths using three-dimensional intelligent scissors. 3D Intelligent Scissors has the ability to extract hollow vessels by climbing along surfaces. Thus, a rough path can be found if the end points of a narrow tunnel are selected. Another advantage of using 3D Intelligent Scissors is that it can successfully extract the path even the actual path is broken. Very often the diameter of a blood vessel may be sometimes smaller than one voxel. Since our algorithm is semi-automatic, it provides sufficient control to accurately find out the desired paths with only minimal user intervention. We have applied the proposed 3D Intelligent Scissors for the application of tracking lung air-ways. A 3D stylus and cutting planes in the virtual volume space are used for selecting the end nodes of vessel paths and revealing the local details on the texture-mapped plane. Our system also provides advanced navigation features to guide visual examination and diagnosis of volumetric air-way paths. These include interactive control of incremental forward, backward, and rotating navigation in reference to the viewing frustum and 3D real-time input.

Using the Markov random process, we developed two new approaches to pattern recognition: (1) Hidden Markov model for modeling spectral features for recognizing 2D shapes. This is because Fourier spectra are suitable for describing 2D shapes of simple closed contours and probabilistic models are capable of coping with random variations in object shapes. We will analyze the properties of spectral features derived from contours of 2D shapes and use these features in 2D pattern recognition. (2) Markov random fields for modeling 2D structural and statistical features. We will give a theoretic analysis of this approach, discuss the issues in the design of neighborhood system and cliques for Markov random field models, and analyze the properties of the models.

We have applied the proposed approach to the recognition of unconstrained handwritten numerals and 2D shapes. Our extensive experimental results show that the proposed approach can achieve a higher performance than that reported recently in the literature.

Water surface is an important element for rendering a realistic landscape image, particularly for visual environmental assessment. The incident light from the water surface to the viewpoint generally consists of two components, reflected and refracted light. The source of the latter is further subdivided into reflected light from the bottom of the water and scattered light in it. In order to render a realistic water surface, the calculation for each component of the light is a quite complex and time consuming task. We propose the fast rendering algorithms of the water surface for visual environmental assessment of outdoor scenes. The features are: (1) the reflected light from the water surface is rendered very quickly through the quasi-ray tracing techniques, which employ inversion and reflection mapping with consideration of the depth of objects mirrored on the water surface; (2) the refracted light coming through the water surface, which gives the color of water, is efficiently calculated by using a look-up table in which both characteristics of the light sources (both direct sunlight and skylight) and the effects of scattering and attenuation depending on turbidity are taken into account. In order to testify the proposed algorithms, some examples are given.

Reconciling scene realism with interactivity has emerged as one of the most important areas in making virtual reality feasible for large-scale CAD data sets consisting of several millions of primitives. Level of detail (LoD) and multi-resolution modeling techniques in virtual reality can be used to speed up the process of virtual design and virtual prototyping. In this paper, we present an automatic LoD generation and rendering algorithm, which is suitable for CAD models and propose a new multi-resolution representation scheme called MRM (multi-resolution model), which can support efficient extraction of fixed resolution and variable resolution for multiple objects in the same scene. MRM scheme supports unified selective simplifications and selective refinements over the mesh. Furthermore, LoD and multi-resolution models may be used to support real-time geometric transmission in collaborative virtual design and prototyping.

Document image processing has become an increasingly important technology in the automation of office documentation tasks. Automatic document scanners such as text readers and OCR (Optical Character Recognition) systems are an essential component of systems capable of those tasks. One of the problems in this field is that the document to be read is not always placed correctly on a flat-bed scanner. This means that the document may be skewed on the scanner bed, resulting in a skewed image. This skew has a detrimental effect on document analysis, document understanding, and character segmentation and recognition. Consequently, detecting the skew of a document image and correcting it are important issues in realizing a practical document reader. This paper presents the use of analyzing the connected components extracted from the binary image of a document page. Such an analysis provides a lot of useful information, and will be used to perform skew correction, segmentation and classification of the document. Moreover, we describe two new algorithms — one for skew detection and one for skew correction. The new skew correction algorithm we propose has been shown to be fast and accurate, with run times averaging under 1.5 CPU seconds and 30 seconds real time to calculate the angle on a 5000/20 DEC workstation. Experiments on over 100 pages show that the method works well on a wide variety of layouts, including sparse textual regions, mixed fonts, multiple columns, and even for documents with a high graphical content.

In this paper, a new, geometric approach to pattern identification in data mining is presented. It is based on applying string edit distance computation to measuring the similarity between multi-dimensional curves. The string edit distance computation is extended to allow the possibility of using strings, where each element is a vector rather than just a symbol. We discuss an approach for representing 3D-curves using the curvature and the tension as their symbolic representation. This transformation preserves all the information contained in the original 3D-curve. We validate this approach through experiments using synthetic and digitalized data. In particular, the proposed approach is suitable to measure the similarity of 3D-curves invariant under translation, rotation, and scaling. It also can be applied for partial curve matching.