Kye Kyung Kim

Scene text extraction in natural scene images using hierarchical feature combining and verification

We propose a method that extracts text regions in natural scene images using low-level image features and that verifies the extracted regions through a high-level text stroke feature. Then the two level features are combined hierarchically. The low-level features are color continuity, gray-level variation and color variance. The color continuity is used since most of the characters in a text region have the same color, and the gray-level variation is used since the text strokes are distinctive to the background in their gray-level values. Also, the color variance is used since the text strokes are distinctive in their colors to the background, and this value is more sensitive than the gray-level variations. As a high level feature, text stroke is examined using multi-resolution wavelet transforms on local image areas and the feature vector is input to a SVM (support vector machine) for verification. We tested the proposed method with various kinds of the natural scene images and confirmed that extraction rates are high even in complex images.

Segmentation-based recognition of handwritten touching pairs of digits using structural features

In this paper, we propose a segmentation-based recognition method of handwritten touching pairs of digits using structural features of contour. Four kinds of candidate break points are obtained from contour and six touching types are defined based on an analysis of the ligature and the characteristics of candidate break points. The final break points of touching pairs of digits are deduced by verifying candidate segment combinations. The main advantages of this method are that reliable segment combinations are used in the multiple hypothesis recognition, and segmentation error of traditional segmentation-based recognition method are reduced by verifying segment combinations. To evaluate the proposed method, we have experimented with 3500 touching pairs of digits of the NIST SD19 database. An encouraging recognition rate of 92.5% has been obtained.

A Robot Photographer with User Interactivity

This paper describes a new robot photographer system which can interact with people. The goal of this research is to make the system act like a human photographer. This system is based on a mobile robot having capabilities of wireless communication and stereo vision. It recognizes waving hands of people, moves toward them, and takes pictures with designated compositions. The pictures are transmitted to personal computers over the wireless network. In comparison with previous researches, the most unique things of this system are human interaction and user preference. To realize these properties, this new robot photographer system is considered optical property of lens and applied interesting vision algorithm which was never tried before for previous robot photographer systems

Object Recognition Method for Industrial Intelligent Robot

The introduction of industrial intelligent robot using vision sensor has been interested in automated factory. 2D and 3D vision sensors have used to recognize object and to estimate object pose, which is for packaging parts onto a complete whole. But it is not trivial task due to illumination and various types of objects. Object image has distorted due to illumination that has caused low reliability in recognition. In this paper, recognition method of complex shape object has been proposed. An accurate object region has detected from combined binary image, which has achieved using DoG filter and local adaptive binarization. The object has recognized using neural network, which is trained with sub-divided object class according to object type and rotation angle. Predefined shape model of object and maximal slope have used to estimate the pose of object. The performance has evaluated on ETRI database and recognition rate of 96% has obtained.

Interaction control for postural correction on a riding simulation system

A horseback riding simulator is a robotic machine that simulates the motion of horseback riding. In this paper, we present an interaction control system for a horseback riding simulator. The proposed system provides a postural correction function suited to the user based on their historical log and specialized posture coaching data. The system has adopted certain schemes for posture detection and recognition of the identified user as a way to recommend customized exercise modes. Our experiments show that including these techniques will help users maintain good posture while riding. Categories and Subject Descriptors J.7 [Computer Applications]: Computers in Other Systems; H.5.m [Information Interfaces and Presentation]: Miscellaneous

Human age estimation using multi-class SVM

Age estimation from face images has attracted attention because it is expected to have many application fields and growing interest. Human age estimation is very difficult tasks because a person has a different in appearance, which varies along with environment even same age. And also, pose, lighting condition or expression has an effect to estimate human age. Age estimation has challenged due to aforementioned problem even it has various potential application fields. In this paper, age estimation using Gabor feature and support vector machine as a classifier has proposed. Age-specific face images has saved in database, which has captured in real world environment. Age estimation result has applied to interact with sports simulator, which provides specialized information to each person, who wants to get individualized exercise model on sports simulator. We have evaluated age estimation performance on ETRI database, which has constructed during several months in real world environment.

Real-time user pose verification in a depth image for simulator

We propose a user pose verification technique using depth information to compare their pose with expert riders. Xtion sensor by Asus is used for gathering a depth data in real world. The user pose verification algorithm is divided into two categories: user segmentation and user pose verification. In user segmentation step, body parts are segmented based on the region growing algorithm from a head point (i.e. seed point). Then, a simple algorithm is used to generate skeletal joints in the segmented body parts. Finally, the user pose is investigated by the standard pose of experts in the same situation.

Multiple objects recognition for industrial robot applications

Vision-based object recognition has been studied intensively because of many application fields, especially, manufacturing process in industrial robot application. But it has been challenged due to illumination effect, diverse material object, atypical shape object, etc. In this paper, multiple object recognition including complex shape object has been proposed. The object is consisted of variable characteristic, which has reflection material surface wrapped by plastic or flexible shape. Object segmentation using back light and pose estimation by maximal axis detection, and object recognition by NN have developed. We have evaluated recognition performance on database of ETRI, which has acquired under various lighting conditions.

Precise pose estimation using landmark feature extraction and blob analysis for bin picking

This paper proposes a method of vision based pose estimation of randomly piled objects. It is necessary to estimate precise rotation angle of picking objects. However, it is non-trivial task because an object placed in every position makes distorted image far from right position image. We propose a precise pose estimation method of bin picking objects. The landmark feature of a picking object is extracted and is used to estimate coarse rotation angle of object. The pose of the object can be precisely estimated by analyzing feature blobs in the landmark. We have experimented to evaluate proposed method in the pilot system, which is constructed like a real manufacturing environment. The experiments have been conducted using 1,000 images. As a result, the pose estimation accuracy of less than 0.154 degree has obtained.

Multiple concurrent operations and flexible robotic picking for manufacturing process environments

For automation of part-packaging processes, the system should be able to perform certain functions for operating the applicable concurrent pickup of randomized parts stored in multiple containers. To achieve such operation, we present an effective vision based picking system for multiple concurrent tasks in small parts manufacturing sites. The proposed system detects target objects and analyzes whether they are available for a concurrent picking operation using dual-arm or multiple arm manipulators. In the experimental study, the method showed improved accuracy and an increase in the number of target objects compared to the simple area limitation method for the collision avoidance between the manipulator. The system can be a solution for multiple concurrent operation and flexible robotic picking in manufacturing process environments.