• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.22 no.3_1spc
• /
• pp.496-503
• /
• 2013

Recently, the death rate owing to breast cancers has been increasing, and the occurrence age for breast cancers is lowering every year. Mammography is known to be a reliable detection method for breast cancers and works by detecting texture changes, calcifications, and other potential symptoms. In this research on breast cancer detection, candidate objects were detected by using image processing on mammograms, and feature analysis was used to classify candidate objects as benign tumors and malignant tumors. To find candidate objects, image pre-processing and binarization using multiple thresholds, and the grouping of micro-calcifications were used. More than 50 shape features and intensity features were used in the classification. The performance of the detection algorithm by using Euclidian distance method for benign tumors was 93%, and the classification error rate was approximately 2%.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.53 no.3
• /
• pp.91-98
• /
• 2016

Detection and classification of underwater objects in sonar imagery are challenging problems. This paper proposes a system that detects and identifies underwater objects at the sea floor level using a sonar image and image processing techniques. The identification process of underwater objects consists of two steps; detection of candidate regions and identification of underwater objects. The candidate regions of underwater objects are extracted by image registration through the detection of common feature points between the reference background image and the current scanning image. And then, underwater objects are identified as the closest pattern within the database using eigenvectors and eigenvalues as features. The proposed system is expected to be used in efficient securement of Q route in vessel navigation.

• Journal of the Institute of Convergence Signal Processing
• /
• v.11 no.3
• /
• pp.203-208
• /
• 2010

In this paper, we convert underwater acoustic signal made by sonar system into digital image. We propose the algorithm that detects target candidate and emphasizes information of target introducing image processing technique for the digital image. The process detecting underwater target estimates background noise in underwater acoustic signal changing irregularly, recomposes it. and eliminates background from original image. Therefore, it generates initial target group. Also, it generates weighted map through proceeding doppler information, ensures information for target candidate through filtering using weighted map for image eliminated background noise, and decides the target candidate area in the single frame. In this paper, we verified that proposed algorithm almost had eliminated the noise generated irregularly in underwater acoustic signal made by simulation, targets had been displayed more surely in the image of underwater acoustic signal through filtering and process of target detection.

• Proceedings of the KSRS Conference
• /
• 2006.03a
• /
• pp.7-14
• /
• 2006

Detection and identification of targets from remotely sensed imagery are of great interest for civilian and military application. This paper presents an algorithm for target detection in high spatial resolution imagery based on the spectral and the dimensional characteristics of the reference target. In this algorithm, the spectral and the dimensional information of the reference target is extracted automatically from the sample image of the reference target. Then in the entire image, the candidate target pixels are extracted based on the spectral characteristics of the reference target. Finally, groups of candidate pixels which form isolated spatial objects of similar size to that of the reference target are extracted as detected targets. The experimental test results showed that even though the algorithm detected spatial objects which has different shape as targets if the spectral and the dimensional characteristics are similar to that of the reference target, it could detect 97.5% of the targets in the image. Using hyperspectral image and utilizing the shape information are expected to increase the performance of the proposed algorithm.

• Korean Journal of Remote Sensing
• /
• v.22 no.1
• /
• pp.41-47
• /
• 2006

Detection and identification of targets from remotely sensed imagery are of great interest for civilian and military application. This paper presents an algorithm for target detection in high spatial resolution imagery based on the spectral and the dimensional characteristics of the reference target. In this algorithm, the spectral and the dimensional information of the reference target is extracted automatically from the sample image of the reference target. Then in the entire image, the candidate target pixels are extracted based on the spectral characteristics of the reference target. Finally, groups of candidate pixels which form isolated spatial objects of similar size to that of the reference target are extracted as detected targets. The experimental test results showed that even though the algorithm detected spatial objects which has different shape as targets if the spectral and the dimensional characteristics are similar to that of the reference target, it could detect 97.5% of the targets in the image. Using hyperspectral image and utilizing the shape information are expected to increase the performance of the proposed algorithm.

• Journal of the Korea Society of Computer and Information
• /
• v.5 no.2
• /
• pp.22-29
• /
• 2000

The purpose of this paper is to propose a method to maximize a content-based image retrieval efficiency in multiple images. To perform an image retrieval job efficiently, it is necessary to minimize the number of candidate-images. Furthermore, a miximum efficiency of image retrieval could not be expected if an image retrieval job in the multiple images is done on the basis of patterns of single image distinctive features. In this method, a multiple SOFM neural network system is adopted to select automatically distinctive feature patterns which have a maximum efficiency of image retrieval in the multiple images. In this method. an image retrieval efficiency is improved 3% than individual features and the number of candidate-images is reduced by the multiple SOFM neural network system.

• Journal of KIISE:Software and Applications
• /
• v.37 no.4
• /
• pp.264-272
• /
• 2010

Example-based super resolution is a method to restore a high resolution image from low resolution images through training and retrieval of image patches. It is not only good in its performance but also available for a single frame low-resolution image. However, its time complexity is very high because it requires lots of comparisons to retrieve image patches in restoration process. In order to improve the restoration speed, an efficient patch retrieval algorithm is essential. In this paper, we applied various high-dimensional feature retrieval methods, available for the patch retrieval, to a practical example-based super resolution system and compared their speed. As well, we propose to apply the multi-phase candidate reduction approach to the patch retrieval process, which was successfully applied in character recognition fields but not used for the super resolution. In the experiments, LSH was the fastest among conventional methods. The multi-phase candidate reduction method, proposed in this paper, was even faster than LSH: For $1024{\times}1024$ images, it was 3.12 times faster than LSH.

• Journal of Korea Multimedia Society
• /
• v.18 no.2
• /
• pp.138-148
• /
• 2015

Object detection methods based on background learning are widely used in video surveillance. However, when a car runs with headlights on, these methods are likely to detect the car region and the area illuminated by the headlights as one connected change region. This paper describes a method of separating the car region from the area illuminated by the headlights. First, we detect change regions with a background learning method, and extract blobs, connected components in the detected change region. If a blob is larger than the maximum object size, we extract candidate object regions from the blob by clustering the intensity histogram of the frame difference between the mean of background images and an input image. Finally, we compute the similarity between the mean of background images and the input image within each candidate region and select a candidate region with weak similarity as an object region.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.20 no.2
• /
• pp.262-266
• /
• 2010

Recognizing a license plate of a vehicle has widely been issued. In this thesis, firstly, mean shift algorithm is used to filter and segment a color vehicle image in order to get candidate regions. These candidate regions are then analyzed and classified in order to decide whether a candidate region contains a license plate. We then present an approach to recognize a vehicle's license plate using the Fuzzy ARTMAP neural network, a relatively new architecture of the neural network family. We show that the proposed system is well to recognize the license plate and shows some compute simulations.

• Korean Journal of Remote Sensing
• /
• v.11 no.1
• /
• pp.47-64
• /
• 1995

In this paper, an algorithm that finds man-made structures in a praylevel aerial images is proposed to perform stereo matching. An extracted contour of buildings must have a high accuracy in order to get a good feature-based stereo matching result. Therefore this study focuses on the use of edge following in the original image rather than use of ordinary edge filters. The Algorithm is composed of two main categories; one is to find candidate regions in the whole image and the other is to extract exact contours of each building which each candidate region.. The region growing method using the centroid linkage method of variance value is used to find candidate regions of building and the contour line tracing algorithm based on an adge following method is used to extract exact contours. The result shows that the almost contours of building composed of line segments are extracted.