Young Kug Ham

Recognition of human front faces using knowledge-based feature extraction and neurofuzzy algorithm

Abstract A recognition method of human front faces using knowledge-based feature extraction and a neuro-fuzzy algorithm is proposed. In the preprocessing step we extract the face part from the homogeneous background by tracking face boundaries, where we assume that the face part is located in the center of a captured image. Then, based on a priori knowledge of human faces, we extract five normalized features. In the recognition step we propose a neuro-fuzzy algorithm that employs a trapezoidal fuzzy membership function and modified error backpropagation (EBP) algorithm. The former absorbs variation of feature values and the latter shows good learning efficiency. Computer simulation results with 80 test images of 20 persons show that the proposed neuro-fuzzy method yields higher recognition rate than the conventional ones.

3D object recognition in range images using hidden markov models and neural networks

Abstract This paper proposes an effective algorithm for 3D object recognition in range images using hidden Markov models (HMMs) and neural networks (NNs). Because an HMM trains each object independently, it is adequate to recognition applications in which a modelbase is frequently updated. Also NNs are employed as a postprocessing step to increase the recognition rate. To avoid the dependency of feature values on the viewing position, we employ 3D features such as surface type, moments, surface area, line length, and relationship between adjacent surfaces. In representing feature values, a fuzzy membership function is used to absorb variation of feature values caused by occlusion of the 3D object. Computer simulation results for synthetic and real range images show that the proposed HMM-based system combined with NNs can be successfully applied to recognition of 3D objects.

Hybrid approaches to frontal view face recognition using the hidden Markov model and neural network

Abstract In this paper, for frontal view face recognition a hidden Markov model (HMM) algorithm and hybrid approaches using the HMM and neural network (NN) are proposed. In the preprocessing stage, edges of a face are detected using the conventional locally adaptive threshold (LAT) scheme and facial features are extracted based on generic knowledge of facial components. In constructing a database with normalized features, we employ HMM parameters of each person computed by the forward-backward algorithm. Computer simulation shows that the proposed HMM-NN algorithm yields higher recognition rate compared with several conventional face recognition algorithms.

Recognition of raised characters for automatic classification of rubber tires

The recognition of raised alphanumeric markings on rubber tires for their automatic classification is presented. Raised alphanumeric markings on rubber tires show different characteristics from those of printed characters. In the preprocessing step of the proposed method, we first determine the slope of an arc, along which alphanumerics are marked, using the Hough transform, and align them horizontally. Then we separate each character using vertical and horizontal projections. In the recognition step, to recognize characters hierarchically we use several effective features, such as width of a character, number of cross points, partial projections, and distance features. Computer simulation results show that the proposed system can be successfully applied to automatic classification of rubber tires.

On-line recognition of cursive Korean characters using DP matching and fuzzy concept

Abstract This paper proposes an on-line recognition method of cursive Korean characters based on dynamic programming (DP) matching and fuzzy concept. The proposed algorithm, invariant to rotation and size, reduces greatly the computational requirement of DP matching by matching phonemes rather than character patterns, where the angle difference and the ratio of lengths between input and reference patterns are adopted as matching features. Correct matching of poorly-written cursive characters becomes possible by introducing the fuzzy concept in representing phoneme features and the positional relationships between adjacent phonemes. Computer simulation results show the effectiveness of the proposed algorithm.

Automated analysis of mixed documents consisting of printed Korean/alphanumeric texts and graphic images

An efficient algorithm is proposed that recognizes a mixed document consisting of printed Korean/alphanumeric text and graphic images. In the preprocessing step, an input document is skew-normalized, if necessary, by rotating it by an angle detected with the Hough transform. Then we separate the graphic image parts from the text parts by considering chain codes of connected components. We further separate each character using vertical and horizontal projections. In the recognition step, a mixed text consisting of two different sets of characters, e.g. , Korean and alphanumeric characters is recognized. Korean and alphanumeric characters are classified and each is recognized hierarchically using several effective features. The output is obtained by combining the recognized characters and separated graphic parts. An efficient automated analysis algorithm for mixed documents consisting of graphic images and two different sets of characters is proposed and its performance is demonstrated via computer simulation.

Knowledge-based face recognition using neural networks

In this paper, we propose a method for facial feature extraction and recognition algorithm based on neural networks. First we separate the face part from the captured image based on the fact that the face image is located in the center of an input image and the background is relatively uniform. Then we obtain 4 normalized features from the extracted face image. For face recognition, we use the backpropagation technique of the neural networks. The proposed knowledge-based technique recognizes 14 persons correctly.

Fuzzy-based recognition of human front faces using the trapezoidal membership function

A fuzzy-based recognition method of human front faces using the trapezoidal membership function is proposed. In the preprocessing step, we extract the face part from the background image by tracking face boundaries under the assumption that the face part is located in the center of a captured image with homogeneous background. Then based on the a priori knowledge of human faces we extract five normalized features. In the recognition step, we propose a fuzzy-based algorithm that employs a trapezoidal membership function that absorbs the variation of feature values of the same person. Computer simulation results with 80 test images of 20 persons show that the proposed method yields higher recognition rate than the conventional ones.<<ETX>>

Simple sequentially designed rule-based alphanumerics recognition algorithm for OCR document processing using a thinning process

A simple method to recognize the printed alphanumerics is discussed. The proposed method is a simple rule-based structural method to recognize printed alphanumerics of image scanner data based on the thinning operation. This paper also presents major achievement made toward the development of a fast hierarchical recognition scheme for the printed and handwritten facsimile data. The conventional thinning techniques give good results for high-resolution image scanner data, but they suffer drawbacks for low-resolution data. Our scheme recognizes 55 characters per second on the IBM PC/386 environment and the recognition rate is 98%.

Analysis of Mixed Korean Documents Using the Branch and Bound Algorithm Based on DP Matching

This paper presents an effective automated analysis system for mixed documents consisting of handwritten texts and graphic images. In the preprocessing step, an input image is binarized, then graphic regions are separated from text parts using chain codes of connected components. In the character recognition step, we recognize two different sets of handwritten characters: Korean and alphanumeric characters. Considering the structural complexity and variations of Korean characters, we separate them based on partial recognition results of vowels and extract primitive phonemes using a branch and bound algorithm based on dynamic programming (DP) matching. Finally, to validate recognition results, a dictionary and knowledge are employed. Computer simulation with 50 test documents shows that the proposed algorithm analyzes effectively mixed documents.