Ku-Jin Kim

Towards valid parametric CAD models

In variational CAD design, parametric models may fail to regenerate raising the question of which parameter values lead to valid models. The problem is easy to state but difficult to solve. Using a simplification, we present an algorithm that computes for parametric models valid parameter ranges within which the model will regenerate. We explain also why the general problem is hard.

Vehicle Color Classification Based on the Support Vector Machine Method

We present a vehicle color classification method from outdoor vehicle images. Although the vehicle color recognition is important especially for the newest applications including ITS (intelligent transportation system), we have no significant previous results at least to our knowledge. In this paper, we started from converting the vehicle image into an HSV(hue-saturation-value) color model-based image, to eliminate distortions due to the intensity changes. Then, we construct the feature vector, which is a two-dimensional histogram for the hue and saturation pairs. We use the SVM(support vector machine) method to classify these feature vectors into five vehicle color classes: black, white, red, yellow and blue. Our implementation result shows 94.92% of success rate for 500 outdoor vehicle images.

The Perspective Silhouette of a Canal Surface

We present an efficient and robust algorithm for parameterizing the perspective silhouette of a canal surface and detecting each connected component of the silhouette. A canal surface is the envelope of a moving sphere with varying radius, defined by the trajectory C(t) of its center and a radius function r(t) . This moving sphere, S(t) , touches the canal surface at a characteristic circle K(t) . We decompose the canal surface into a set of characteristic circles, compute the silhouette points on each characteristic circle, and then parameterize the silhouette curve. The perspective silhouette of the sphere S(t) from a given viewpoint consists of a circle Q(t) ; by identifying the values of t at which K(t) and Q(t) touch, we can find all the connected components of the silhouette curve of the canal surface.

Minimum distance between a canal surface and a simple surface

Abstract The computation of the minimum distance between two objects is an important problem in the applications such as haptic rendering, CAD/CAM, NC verification, robotics and computer graphics. This paper presents a method to compute the minimum distance between a canal surface and a simple surface (i.e. a plane, a natural quadric, or a torus) by finding roots of a function of a single parameter. We utilize the fact that the normals at the closest points between two surfaces are collinear. Given the spine curve C(t), tmin≤t≤tmax, and the radius function r(t) for a canal surface, a point on the spine curve C(t ∗ ) uniquely determines a characteristic circle K(t ∗ ) on the surface. Normals to the canal surface at points on K(t ∗ ) form a cone with a vertex C(t ∗ ) and an axis which is parallel to C′(t ∗ ). Then we construct a function of t which expresses the condition that the perpendicular from C(t) to a given simple surface is embedded in the cone of normals to the canal surface at points on K(t). By solving this equation, we find characteristic circles which contain the points of locally minimum distance from the simple surface. Based on these circles, we can compute the minimum distance between given surfaces.

Kinematic analysis of spatial fixed-axis higher pairs using configuration spaces

We present a kinematic analysis algorithm for spatial higher pairs whose parts rotate around or translate along fixed spatial axes. The part geometry is specified in a parametric boundary representation consisting of planar, cylindrical, and spherical patches bounded by line and circle segments. Kinematic analysis is performed by configuration space construction following the method that we developed for planar pairs. The configuration space of a pair is a complete encoding of its kinematics, including contact constraints, contact changes, and part motions. The algorithm constructs contact curves for all pairs of part features, computes the induced configuration space partition, and identifies the free space components. Spatial contact analysis is far harder than planar analysis because there are 72 types of contact versus 8. We have developed a systematic analysis technique and have used it to derive low-degree equations for all cases, which are readily solvable in closed form or numerically. We demonstrate the implemented algorithm on three design scenarios involving spatial pairs and planar pairs with axis misalignment.

Deciding the Number of Color Histogram Bins for Vehicle Color Recognition

Given vehicle images, we suggest a way to recognize the color of the vehicle contained in the image. The color feature of a vehicle is represented by a color histogram, and we decide the appropriate number of color histogram bins, which mainly affects the successful recognition rate. After generating the histograms, template matching is used to decide the vehicle color. In HSI (hue saturation intensity) color space, experimental results show that the partition of H, S, and I into 8, 4, 4, respectively, achieves the highest success rate up to 88.34%.

Retrieval of Identical Clothing Images Based on Local Color Histograms

In this paper, we present a novel approach to retrieve the person images that contain the identical clothing to a query image from the image set captured by multiple CCTV cameras. In order to measure the similarity of the clothing, we analyze the color information of the clothing area in the image. To find the clothing area, we detect the face area first. The clothing area is found based on the position of the face area. Then, we apply the color quantization to the clothing area. We build six color histograms based on the quantized color for six sub-regions defined in the clothing area. The feature vector for the clothing area is composed by using the color histograms. Similarity between two clothing areas is measured by Euclidean distance between the feature vectors. In the experimental results, our approach shows the better performance compared with the method that uses HSV histogram-based color analysis.

Torus/Sphere intersection based on a configuration space approach

Abstract This paper presents an efficient and robust geometric algorithm that classifies and detects all possible types of torus/sphere intersections, including all degenerate conic sections (circles) and singular intersections. Given a torus and a sphere, we treat one surface as an obstacle and the other surface as the envelope surface of a moving ball. In this case, the Configuration space ( C-space ) obstacle is the same as the constant radius offset of the original obstacle, where the radius of the moving ball is taken as the offset distance. Based on the intersection between the C-space obstacle and the trajectory of the center of the moving ball, we detect all the intersection loops and singular contact point/circle of the original torus and sphere. Moreover, we generate exactly one starting point (for numerical curve tracing) on each connected component of the intersection curve. All required computations involve vector/distance computations and circle/circle intersections, which can be implemented efficiently and robustly. All degenerate conic sections (circles) can also be detected using a few additional simple geometric tests. The intersection curve itself (a quartic space curve, in general) is then approximated with a sequence of cubic curve segments.

Shrinking: another method for surface reconstruction

We present a method to reconstruct a pipe or a canal surface from a point cloud (a set of unorganized points). A pipe surface is defined by a spine curve and a constant radius of a swept sphere, while a variable radius may be used to define a canal surface. In this paper, by using the shrinking and moving least-squares methods, we reduce a point cloud to a thin curve-like point set which will be approximated to the spine curve of a pipe or canal surface. The distance between a point in the thin point cloud and a corresponding point in the original point set represents the radius of the pipe or canal surface.

Color recognition with compact color features

For color images, color histograms are generally used as the color feature vectors for classifying the colors of objects. To achieve a higher success rate in color classification, feature vectors with a higher dimension are required, yet this causes a low efficiency with regard to the computation time and memory usage. Therefore, this paper proposes a method of reducing the feature vector dimension by a factor of 170 based on combining two techniques: (i) projecting a color histogram generated in 3D color space into 2D color planes and (ii) converting the color histograms to class histograms using a naive Bayesian classifier. The resulting feature vectors are then classified using a support vector machine method and template matching method to recognize the object colors. With both classification methods, a better recognition rate is achieved than when using the original large feature vectors. Copyright