Junsik Lim

Automatic detection and recognition of Korean text in outdoor signboard images

In this paper, an automatic translation system for Korean signboard images is described. The system includes detection and extraction of text for the recognition and translation of shop names into English. It deals with impediments caused by different font styles and font sizes, as well as illumination changes and noise effects. Firstly, the text region is extracted by an edge-histogram, and the text is binarized by clustering. Secondly, the extracted text is divided into individual characters, which are recognized by using a minimum distance classifier. A shape-based statistical feature is adopted, which is adequate for Korean character recognition, and candidates of the recognition results are generated for each character. The final translation step incorporates the database of shop names, to obtain the most probable result from the list of candidates. The system has been implemented in a mobile phone and is demonstrated to show acceptable performance.

Segmentation of Brain MR Images Using an Ant Colony Optimization Algorithm

In this paper, we describe a segmentation method for brain MR images using an ant colony optimization (ACO) algorithm. This is a relatively new meta-heuristic algorithm and a successful paradigm of all the algorithms which take advantage of the insect’s behavior. It has been applied to solve many optimization problems with good discretion, parallel, robustness and positive feedback. As an advanced optimization algorithm, only recently, researchers began to apply ACO to image processing tasks. Hence, we segment the MR brain image using ant colony optimization algorithm. Compared to traditional meta-heuristic segmentation methods, the proposed method has advantages that it can effectively segment the fine details.

Automatic Detection and Recognition of Shop Name in Outdoor Signboard Images

In this paper, a system for automatic detection and recognition of Korean texts or shop names in outdoor signboard images is described. The system includes detection, binarization and extraction of text in a signboard image captured by a camera of a mobile phone for the recognition of the shop name. It can deal with different font styles and sizes as well as illumination changes. Individual characters detected by connected component analysis are recognized by using nonlinear mesh, in which feature vectors of vertical and horizontal components are extracted from the binarized image. Proposed methods have been applied to a Korean text translation system, which can automatically detect and recognize Korean texts and generate the translation result.

Recognition of Text in Wine Label Images

In this paper, an automatic recognition system for Wine label images is described. The system includes detection and extraction of text for the recognition for Wine label images. It deals with impediments caused by different font styles and font sizes, as well as illumination changes and noise effects. Firstly, the text region is extracted by an edge-histogram, and the text is binarized by clustering. Secondly, the extracted text is divided into individual characters, which are recognized by using the Multi-Layer Perceptron. A shape-based statistical feature is adopted and the recognition results are generated for each character. The system has been implemented in a mobile phone and is demonstrated to show an acceptable performance.

Segmentation and visualization of anatomical structures from volumetric medical images

This paper presents a method that can extract and visualize anatomical structures from volumetric medical images by using a 3D level set segmentation method and a hybrid volume rendering technique. First, the segmentation using the level set method was conducted through a surface evolution framework based on the geometric variation principle. This approach addresses the topological changes in the deformable surface by using the geometric integral measures and level set theory. These integral measures contain a robust alignment term, an active region term, and a mean curvature term. By using the level set method with a new hybrid speed function derived from the geometric integral measures, the accurate deformable surface can be extracted from a volumetric medical data set. Second, we employed a hybrid volume rendering approach to visualize the extracted deformable structures. Our method combines indirect and direct volume rendering techniques. Segmented objects within the data set are rendered locally by surface rendering on an object-by-object basis. Globally, all the results of subsequent object rendering are obtained by direct volume rendering (DVR). Then the two rendered results are finally combined in a merging step. This is especially useful when inner structures should be visualized together with semi-transparent outer parts. This merging step is similar to the focus-plus-context approach known from information visualization. Finally, we verified the accuracy and robustness of the proposed segmentation method for various medical volume images. The volume rendering results of segmented 3D objects show that our proposed method can accurately extract and visualize human organs from various multimodality medical volume images.

Medical image registration using the modified conditional entropy measure combining the spatial and intensity information

We propose an image registration technique using spatial and intensity information. The registration is conducted by the use of a measure based on the entropy of conditional probabilities. To achieve the registration, we first define a modified conditional entropy (MCE) computed from the joint histograms for the area intensities of two given images. In order to combine the spatial information into a traditional registration measure, we use the gradient vector flow field. Then the MCE is computed from the gradient vector flow intensity (GVFI) combining the gradient information and their intensity values of original images. To evaluate the performance of the proposed registration method, we conduct various experiments with our method as well as existing method based on the mutual information (MI) criteria. We evaluate the precision of MI- and MCE-based measurements by comparing the registration obtained from MR images and transformed CT images. The experimental results show that our proposed method is a more accurate technique.

A Generic Framework of Integrating Segmentation and Registration

We propose an integrated framework that can simultaneously segment and register a set of medical images using a geometric active model and a pseudo-likelihood method. First, we segment given medical volume data using a fast matching and geometric deformable model and extract a surface of an object from the segmented volume data. Second, we use the hidden Markov random field model and the pseudo-likelihood method to statistically model the intensity distribution of each voxel at the surface region. We adopt the Bernoulli probability model to formulate a prior distribution of the labeling variable for the transformed voxels. The Gaussian mixture model is taken as a probability distribution function for the intensity of the transformed voxel. We use the deterministic annealing EM (DAEM) algorithm to get the proper estimators for the parameters of the complete-data log likelihood function. Then, we define a new registration measure with the maximization function, called Q-function, obtained by the DAEM algorithm. We evaluate the precision of the proposed approach by comparing the registration traces of our measure with other measures such as Mutual Information or Cross Correlation for the original image and its transformed image with respect to translation and rotation. The experimental results show that our method has great potential power to segment and register various medical images given by different modalities.

Medical Image Registration by Combining Gradient Vector Flow and Conditional Entropy Measure

In this paper, we propose a medical image registration technique combining the gradient vector flow and modified conditional entropy. The registration is conducted by the use of a measure based on the entropy of conditional probabilities. To achieve the registration, we first define a modified conditional entropy (MCE) computed from the joint histograms for the area intensities of two given images. In order to combine the spatial information into a traditional registration measure, we use the gradient vector flow field. Then the MCE is computed from the gradient vector flow intensity (GVFI) combining the gradient information and their intensity values of original images. To evaluate the performance of the proposed registration method, we conduct experiments with our method as well as existing method based on the mutual information (MI) criteria. We evaluate the precision of MI- and MCE-based measurements by comparing the registration obtained from MR images and transformed CT images. The experimental results show that the proposed method is faster and more accurate than other optimization methods.

A Shape-Based Approach to the Registration of Medical Imagery Using Gaussian Mixture Models

The problems of segmentation and registration are traditionally approached separately; yet the accuracy of one is of great importance in influencing the accuracy of the other. We propose a new method, using shape information and a statistical model, to address the problem of multimodality medical image registration. Using the approach presented in this paper, we apply a Q-function to measure the statistical dependence or information redundancy between the probability distributions of corresponding voxels from the region of interest in both images. We define a new registration measure with a Q-function obtained by a Gaussian mixture model (GMM) based on an Expectation-maximization (EM) algorithm. The Q-function is assumed to be maximal if the two images for the registration are geometrically aligned. Using the registration traces based on the Q-function, we evaluate the precision of the proposed approach between MR images and CT images. The experimental results show that our method can be very successful in registering various medical images that use different modalities.

Recognition of Korean Text in Outdoor Signboard Images Using Directional Feature and Fisher Measure

In this paper, we propose a Korean character recognition method from outboard signboard images. We have chosen 808 classes of Korean characters by an analysis of frequencies of appearance in a dictionary of signboard names. The proposed method mainly consists of three steps: feature extraction, rough classification, and coarse classification. The first step is to extract a nonlinear directional segments feature, which is immune to the distortion of character shapes. The second step computes an ordered set of 10 recognition candidates using a minimum distance classifier. The last step reorders the recognition candidates using a Fisher discriminant measure. As experimental results, the recognition accuracy is 80.45% for the first choice, and 93.51% for the top five choices.