Hyun-Deok Kang

Self-localization from the panoramic views for autonomous mobile robots

This paper describes a self-localization method for mobile robots using panoramic view images. A panoramic view image has the information of object locations from the viewer robot perspective and the direction between the objects at each position. Among panoramic view image sequences, the target objects in the image such as traffic signs, building facades, road signs, etc., are located in the real world so that the location and direction enable robots to localize position and direction from their view. With previously captured panoramic images, the method calculates the distance and direction of the region of interest, corresponds the regions between the sequences, and identifies the real world location. To obtain the region, vertical edge line segments which are located in the buildings and traffic sign stems and region segments of monotone color are utilized. In the segmented region, a centroid is used to localize a principal point. In our experiments, a mobile robot circulates to capture images of the University campus so that he can recognize his location and make a map of the campus. Experimental results showed the method is adequate for map generation and sending a robot on an errand to a destination.

Object Recognition using Segmented Region and Multiple Features on Outdoor Environments

This paper presents the method of region segmentation like six features, color, edge, straight line, geometric information, and template of tree color. The proposition of method segments region of image which is obtained by CCD camera mounted on the mobile robot. Moving robot takes database images on outdoor environment. We classify object to natural and artificial and then define their characteristics individually. In the process, we segment regions included objects by preprocessing. Objects can be recognized when we combine predefined multiple features. In addition, the feature of XCM (X co-occurrence matrix) detect region of tree, where X is information of arbitrary like intensity or hue. So the features use XCM as well as five features which we define. Our method is more effective than conventional region segmentation on outdoor environment because we present the method to combine various features in complex image. We achieved the result of region segmentation using multiple features through experiments.

Self-localization of autonomous mobile robot from the multiple candidates of landmarks

This paper describes self-localization of a mobile robot from the multiple candidates of landmarks. Our robot uses omni-directional vision system for efficient self-localization. This vision system acquires the visible information of all direction views. Our robot utilizes feature of landmarks that size of feature is bigger than that of others in image such as building, symbol tower, hanging banner etc. Our robot uses vertical edges and those merged region as feature. In our previous work, we found the problem that landmark matching is difficult when selected candidates of landmark belong to region of repeating the vertical edges in image. To overcome these problems, robot uses the merged region of vertical edges. If interval of vertical edges is short then robot bundles them regarding as the same region. Thus, these features are candidates of landmarks. In other words, candidates of landmark are vertical edges and those merged regions. Therefore, the extracted merged region of vertical edge reduces the ambiguity of landmark matching. Robot compares with the candidates of landmark between previous and current image. Then, robot is able to find the same landmark between image sequences using the proposed feature and method. The experiments implemented our campus and achieved the efficient self-localization result using robust landmark matching method.

Reconstruction of Rectangular Plane in 3D Space Using Determination of Non-vertical Lines from Hyperboloidal Projection

This paper describes the 3D reconstruction of planar objects using parallel lines from single panoramic image. Determination of non-vertical lines is depends on position of vanishing points with two lines on panoramic image. The vertical 3D line is projected as radial line on panoramic image and horizontal line is projected as curve or arc in panoramic image. Two parallel vertical lines are converged on center point in calibrated panoramic image. On the contrary, two parallel horizontal lines have the pair of vanishing points on the circle at infinity in panoramic image. We reconstruct the planar objects with parallel lines using the vanishing points and properties of parallelism in lines. Finally, we analysis and present the results of 3D reconstruction of planar objects by synthetic or real image.

Evaluation of feature based image stitching algorithm using OpenCV

Image stitching is an attractive method to merging multiple images. It can produce a wide-angle panoramic photograph while maintaining the quality of the source images. The process is simply performed by overlapping part of the images which contain common scene. Today, panoramic image stitching is widely used in applications such as 360-degree cameras and virtual reality photography. If the stitching technology ensure fast processing speed and accurate performance, it is expected to produce diverse image contents in real time. Moreover, the panoramic images can be mapped to various projective layouts, e.g. spherical and stereographic projection. The mapping of the panoramic image along the suitable projective layout makes it possible to express more stereoscopic space, that is, the space in the photo can give the impression that it looks as if it is in front. It can be utilized when creating 3D reconstruction space with high-resolution panoramic images. This paper describes the concept of feature based image stitching and presents the implementation results. It also provides the results of stitching with more than 10 images and suggests the time-efficient way for stitching multiple images. The time-efficient stitching method can be applied to various applications so that panoramic images can be easily accessed in everyday life.

Building facade detection using geometric planar constraints

This paper describes the approach for detecting the building facade using geometric constrain. A planar building facade is only considered. Most of the building facade has the form of a rectangle, which is produced by two or more vertical and horizontal lines, which are orthogonal to each other on the building. The parallel lines in the 3D space are projected onto lines in the image plane that are grouped along the direction of vanishing points. Line segments are extracted from an image for estimating vanishing points. In order to cluster the line segments corresponding to respective a hypothesized vanishing point, J-linkage algorithm is used for simultaneous estimation of multiple models. After clustering the line segments, vanishing point is estimated by minimizing error between vanishing point and line segments of an image. Building facade is detected by using the relationship between vertical and horizontal vanishing points. Building has a cubic structure with two perpendicular planes, which meet at a line. The intersection points between two horizontal vanishing points are located on the intersection line of two facades. Finally, building facades are extracted by the geometric relationships between vertical vanishing point corresponding to respective horizontal vanishing point.

Kết Hợp Phương Pháp Biểu Diễn Đặc Trưng Cục Bộ Và Kỹ Thuật Random Forests Trong Nhận Dạng Người

Bai bao trinh bay hệ thống phân loại dựa tren kỹ thuật Random forest sử dụng phương phap biểu diễn đặc trưng cục bộ ap dụng trong nhận dạng người. Co hai nội dung chinh được trinh bay trong bai nay để giải quyết vấn đề nhận dạng trong trường hợp cảnh nền thay đổi đa dạng. Thứ nhất, chung toi trinh bay kỹ thuật biểu diễn đặc trưng HOG đa mức độ kich thước vung cục bộ nhằm tăng độ chinh xac của hệ thống phân loại. Phương phap nay cho phep trich rut ra một tập lớn cac đặc trưng, sau đo lọc ra chỉ những phần tử co độ khac biệt cao giữa tập positive va negative dựa vao bộ dữ liệu huấn luyện. Thứ hai, may phân loại sử dụng cấu truc thac nước dựa tren kỹ thuật RF được đề xuất sử dụng để huấn luyện va nhận dạng. Trong trường hợp nay, kỹ thuật decision forest dựa tren việc kết hợp cac quyết định yếu sử dụng nhân phân loại la cac SVMs. Mỗi phân loại yếu sử dụng tập đặc trưng trong một vung cục bộ của mẫu. Hệ thống sử dụng cấu truc thac nước cho phep tăng tốc độ phân loại nhờ vao việc loại bỏ được cac mẫu negatives ma chỉ cần một tập nhỏ đặc trưng cục bộ.

Analysis of Vanishing Curves and Points of Planar Objects from Single Panoramic Image

This paper presents the mapping function of horizontal line segments in panoramic image for 3D reconstruction of planar objects. It is interest to get the 3D geometrical information directly under the single panoramic image. The estimation of the 3D geometric information is to use the several conditions of features such as a vanishing line and point and the geometric relationship of feature as parallelism of lines from single panoramic view. This paper describes the 3D reconstruction about any features using the properties of vanishing points and curves. When the panoramic camera located on ground in perpendicular against the mirror, all vertical line segments are converged into center point in image. Horizontal line with ground plane projected as the curve on the basis of each vanishing point in image; the parallel lines with ground are projected as the curve in panoramic image. Second, it is able to obtain the parallel plane in focal point of mirror using the already known the height of mirror from the ground. The curve of horizontal edge line should intersect with the circle at infinity; two intersection points with pi phase difference converge in pair of vanishing points of each object. Consequently, this paper applies the hyperbolic mirror to get panoramic image and perform the 3D reconstruction. It can be suitable and concrete information of environment so that also robot can know the shape of environment within visible range from single panoramic image.