Yong-Hwan Lee

Content-Based Image Retrieval Using Wavelet Spatial-Color and Gabor Normalized Texture in Multi-resolution Database

Image retrieval is one of the most exciting and fastest growing research areas in the field of multimedia technology. This paper proposes a new efficient image retrieval method that applies a weighted combination of color and texture to the wavelet transform, based on spatial-color and second order statistics, respectively. To evaluate the performance of the proposed algorithm, we assess the simulations performance in terms of average precision and Fscore using several image databases, and perform comparative analysis with existing methods such as MPEG-7. The experimental results show that the proposed approach significantly improves the effectiveness of image retrieval. The proposed descriptor is particularly useful for multi-resolution image search and retrieval.

Facial landmarks detection using improved active shape model on android platform

Detection of facial feature is fundamental for applications such as security, biometrics, 3D face modeling and personal authentication. Active Shape Model (ASM) is one of the most popular local texture models for face detection. This paper presents an issue related to face detection based on ASM, and proposes an efficient extraction algorithm for facial landmarks suitable for use on mobile devices. We modifies the original ASM to improve its performance with three changes; (1) Improving the initialization model using the center of the eyes by using a feature map of color information, (2) Constructing modified model definition and fitting more landmarks than the classical ASM, and (3) Extending and building a 2-D profile model for detecting faces in input image. The proposed method is evaluated on dataset containing over 700 images of faces, and experimental results reveal that the proposed algorithm exhibited a significant improvement of over 10.2xa0% in average success ratio, compared to the classic ASM, clearly outperforming on success rate and computing time.

Wavelet-based Image Denoising with Optimal Filter

Image denoising is basic work for image processing, analysis and computer vision. This paper proposes a novel algorithm based on wavelet threshold for image denoising, which is combined with the linear CLS (Constrained Least Squares) filtering and thresholding methods in the transform domain. We demonstrated through simulations with images contaminated by white Gaussian noise that our scheme exhibits better performance in both PSNR (Peak Signal-to-Noise Ratio) and visual effect.

Emotional Recognition from Facial Expression Analysis Using Bezier Curve Fitting

Extracting and understanding of emotion is of high importance for the interaction among human and machine communication systems. The most expressive way to display the humans emotion is through facial expression analysis. This paper presents and implements an automatic extraction and recognition method of facial expression and emotion from still image. To evaluate the performance of the proposed algorithm, we assess the ratio of success with emotionally expressive facial image database. Experimental results shows average 66% of success to analyze and recognize the facial expression and emotion. The obtained result indicates the good performance and enough to applicable to mobile environments.

Efficient object identification and localization for image retrieval using query-by-region

Localizing an object within an image is a common task in the field of computer vision, and represents the first step towards the solution of the recognition problem. This paper presents an efficient approach to object localization for image retrieval using query-by-region. The new algorithm utilizes correlogram back-projection in the YCbCr chromaticity components to handle the problem of subregion querying. Utilizing similar spatial color information enables users to detect and locate primary location and candidate regions accurately without the need for further information about the number of objects. Comparing this new approach to existing methods, an improvement of 21% was observed in experimental trials. These results reveal that color correlograms are markedly more effective than color histograms for this task.

Implementation of Image Descriptor Based on SURF and DCD

Image retrieval is one of the most exciting and fastest growing research areas in the field of multimedia technology. In this paper, we present and implement an image descriptor that extracts the feature of image using combination of SURF (Speed Up Robust Feature) and DCD (Dominant Color Descriptor) scheme. To evaluate the performance of implemented algorithm, we assess the performance of simulation in terms of average precision on two image database commonly used. The result shows that the proposed approach obtain an enough results which are applicable to mobile environments.

Mobile Image Retrieval Using Integration of Geo-sensing and Visual Descriptor

In this paper, we propose a new efficient photo image retrieval method that automatically indexed for searching relevant photo images using a combination of geo-referenced attributes and low-level visual features. Photo image is labeled with its GPS (Global Positioning System) coordinates at the moment of capture, and a flat layer index is generated with the pair of latitude and longitude. These are then utilized to create a hierarchical layer indexes for spatial information after uploaded to media server. Then, low-level visual features such as color histogram and edge histogram are extracted and combined with geo-spatial information for indexing and searching photo images. For users querying process, the proposed method adopts two different steps as progressive approach, filtering and/or selecting the relevant subset prior to content-based retrieval. To evaluate the performance of the proposed descriptor, we assess the simulation performance in terms of average precision and F-score using digital photo collections. Comparing the proposed approach to search using visual content alone, an improvement of around 20% was observed in experimental trials. These results reveal that combination of context and content analysis is markedly more effective and meaningful than using only visual content for this task.

Advanced face recognition and verification in mobile platforms

Purpose – This paper holds a big advantage to enable to recognize faces, regardless of time and place. Also this provides an independent performance of smart phone, because of its process by a computer of third party not by that of the mobile device. In addition, it is desirable to minimize the expensive operations in mobile device with constraint computational power (i.e. battery consumption). Thus, the authors exclude the process of transmission failed from the input device. The paper aims to discuss these issues. Design/methodology/approach – In this paper, the authors have proposed a new face detection and verification algorithm, based on skin color detection to enable extracting the face region from color images of the mobile phone. And then extracted the facial feature as eigenface, verified whether or not the identity of users is right, applied support vector machine to the region of detected face. Findings – The experimental results for two datasets show that the proposed method achieves slightly ...

Improved Active Shape Model for Efficient Extraction of Facial Feature Points on Mobile Devices

Detection of facial feature is fundamental for applications such as security, biometrics, 3D modeling, and facial expression recognition. Active Shape Model (ASM) is one of the most popular local texture models for face detection. This paper addresses issues related to face detection and implements an efficient extraction algorithm for facial landmarks suitable for use on mobile devices. The original ASM was modified to enhance its performance (1) improving the initialization model using the center of the eyes by utilizing a feature ma of RGB color information, (2) building a modified model definition and fitting more landmarks than the classical ASM, and (3) extending and building a 2-D profile model for detecting faces in input images. The new scheme was evaluated on experimental test set containing over 500 images of faces and found to successfully extract facial features, clearly outperforming the original ASM.

Efficient Image Retrieval Using Advanced Clustering SURF

Image retrieval is one of the most exciting and fastest growing research areas in the field of multimedia technology. In this paper, we propose and implement a new image retrieval method that extracts the feature of image using efficient clustering for SURF (Speed Up Robust Feature) scheme, applicable to mobile environments. Since SURF works only on gray-scale images, our clustering SURF is combined with well-known dominant color descriptor to improve the performance of the system. Also, we calculate the feature vector to remove an unnecessary feature, which is located in the sparse region, not to be clustered in final target features. To evaluate the performance of the proposed algorithm, we assess the simulations performance in terms of average precision on two image databases commonly used. Based on the average precision from all queries, 86% and 79% of all relevant images were retrieved. The result shows that the proposed approach obtains an enough results which are applicable to mobile environments.