• Journal of Broadcast Engineering
• /
• v.7 no.2
• /
• pp.145-152
• /
• 2002

As the number of digital images available to users is exponentially growing due to the rapid development of digital technology, content-based image retrieval (CBIR) has been one of the most active research areas. A variety of image retrieval methods have been proposed, where, given an input query image, the images that are similar to the input are retrieved from an image database based on low-level features such as colors and textures. However, most of the existing retrieval methods did not consider the case when an input query image is a part of a whole image in the database due to the high complexity involved in partial matching. In this paper, we present an efficient method for partial image matching by using the histogram distribution relationships between query image and whole image. The proposed approach consists of two steps: the first step prunes the search space and the second step performs block-based retrieval using partial image matching to rank images in candidate set. The experimental results demonstrate the feasibility of the proposed algorithm after assuming that the response tune of the system is very high while retrieving only by using partial image matching without Pruning the search space.

• Journal of KIISE:Computing Practices and Letters
• /
• v.10 no.1
• /
• pp.57-65
• /
• 2004

In the block matching algorithm, search patterns of different shapes or sizes and the distribution of motion vectors have a large impact on both the searching speed and the image quality. In this paper, we propose a new fast block matching algorithm using the unit-diamond search pattern and the flat-hexagon search pattern. Our algorithm first finds the motion vectors that are close to the center of search window using the unit-diamond search pattern, and then fastly finds the other motion vectors that are not close to the center of search window using the flat-hexagon search pattern. Through experiments, compared with the hexagon-based search algorithm(HEXBS), the proposed unit-diamond and flat-hexagonal pattern search algorithm(UDFHS) improves as high as 11∼51% in terms of average number of search point per motion vector estimation and improves about 0.05∼0.74㏈ in terms of PSNR(Peak Signal to Noise Ratio).

• Journal of Korea Multimedia Society
• /
• v.13 no.9
• /
• pp.1348-1356
• /
• 2010

With the advancement of information and communication technology, the market of Internet-based fashion/coordination shopping malls have been considerably increasing year by year. As the number of these Internet shopping malls increases, the operators of the malls tend to decorate the first page of their websites with a variety of events and samples of the best-fit upper-lower clothing pairs. They try to provide visitors of their web sites with products that can induce fresh impression by modifying the first page on a daily or a few days basis. If pairs of best-fit upper-lower clothes for various products available in online shopping malls can be calculated and marked, it would help not only to make the first page of the malls more appealing but also to enable users to purchase linked products in a more convenient way, replacing the recommendations usually made by offline clerks. In the paper, we present the results of designing and implementing an upper-lower clothes matching system in which expert coordinators register matching-value of upper and lower clothes in the form of attribute-value matrix.

• The KIPS Transactions:PartB
• /
• v.9B no.1
• /
• pp.29-34
• /
• 2002

Fingerprint recognition technology was studied by classification and matching. In general, there are five different classifications left loop, right loop, whore, arch, and tented-arch. These classifications are used to determine which class an individual's fingerprint belong to, thereby identifying the individual's fingerprint pattern. The result of this classification, which is sent to the large fingerprint database as an index, helps reduce the matching time and enhance the accuracy of fingerprint matching. The existing fingerprint classification method relies on the number and location of cores and delta points called singular points. The drawback of this method is the lack of accuracy stemming from the classification difficulty involving unclear and/or partially-erased fingerprints. The current paper presents an efficient classification method to rectify the problem associated with identifying Singular points from unclear fingerprints. This method, which is based on Gabor filter's unique characteristics for magnifying directional patterns and frequency range selections, improves fingerprint classification accuracy significantly. In this paper, this method is described and its test result is presented for verification.

• Journal of Broadcast Engineering
• /
• v.16 no.5
• /
• pp.861-872
• /
• 2011

Many fast block-matching algorithms (BMAs) in motion estimation field reduce computational complexity by screening the number of checking points. Although many fast BMAs reduce computations, sometimes they should endure matching errors in comparison with full-search algorithm (FSA). In this paper, a novel fast BMA for constrained one-bit transform (C1BT)-based motion estimation is proposed in order to decrease the calculations of the block distortion measure. Unlike the classical fast BMAs, the proposed algorithm shows a new approach to reduce computations. It utilizes the binomial distribution based on the characteristic of binary plane which is composed of only two elements: 0 and 1. Experimental results show that the proposed algorithm keeps its peak signal-to-noise ratio (PSNR) performance very close to the FSA-C1BT while the computation complexity is reduced considerably.

• Journal of the Korean Institute of Telematics and Electronics S
• /
• v.35S no.6
• /
• pp.101-108
• /
• 1998

An efficient approach to recognize occluded objects is to detect a number of essential features on the boundary of the unknown shape. The salient points including corner points, tangential points and inflection points are detected by the relation of neighboring pixels of each pixel on the boundaries. Corner points are usually detected in the curvature function and tangential points and inflection points are detected by median filtering the curvature function to avoid the effect of quantization noise as corner points is not sufficient to represent an object with lines and arcs. Then, these salient points are used as features for object matching. Discrete Hopfield Neural Network is used for object matching. Experimental results show that the matching result using salient points is better than those of using corner points only when an object consists of lines and arcs.

• Journal of Sensor Science and Technology
• /
• v.19 no.2
• /
• pp.99-111
• /
• 2010

This paper proposes an efficient hardware architecture to estimate disparities between 2D images for generating 3D depth images in a stereo vision system. Stereo matching methods are classified into global and local methods. The local matching method uses the cost functions based on pixel windows such as SAD(sum of absolute difference), SSD(sum of squared difference) and NCC(normalized cross correlation). The NCC-based cost function is less susceptible to differences in noise and lighting condition between left and right images than the subtraction-based functions such as SAD and SSD, and for this reason, the NCC is preferred to the other functions. However, software-based implementations are not adequate for the NCC-based real-time stereo matching, due to its numerous complex operations. Therefore, we propose a fast pipelined hardware architecture suitable for real-time operations of the NCC function. By adopting a block-based box-filtering scheme to perform NCC operations in parallel, the proposed architecture improves processing speed compared with the previous researches. In this architecture, it takes almost the same number of cycles to process all the pixels, irrespective of the window size. Also, the simulation results show that its disparity estimation has low error rate.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2013.05a
• /
• pp.635-637
• /
• 2013

This paper proposed a new hardware architecture for stereo matching in real time. We minimized the amount of calculation and the number of memory accesses through analyzing calculation of stereo matching. From this, we proposed a new stereo matching calculating cell and a new hardware architecture by expanding it in parallel, which concurrently calculates cost function for all pixels in a search range. After expanding it, we proposed a new hardware architecture to calculate cost function for 2-dimensional region. The implemented hardware can be operated with minimum 250Mhz clock frequence in FPGA environment, and has the performance of 813fps in case of the search range of 64 pixels and the image size of $640{\times}480$.

• The KIPS Transactions:PartB
• /
• v.13B no.6 s.109
• /
• pp.579-584
• /
• 2006

This paper proposes a nonlinear template matching measure, called counting measure, as a signal detection measure that is defined as the number of on pixels in the spot area. It is applied to classify probes for an on-off type DNA chip, where each probe spot is classified as hybridized or not. The counting measure also incorporates the maximum response search method, where the expected signal is obtained by taking the maximum among the measured responses of the various positions and sizes of the spot template. The counting measure was compared to existing signal detection measures such as the normalized correlation and the median for 2390 patient samples tested on the human papiliomavirus (HPV) DNA chip. The counting measure performed the best regardless of whether or not the maximum response search method was used. The experimental results showed that the counting measure combined with the positional search was the most preferable.

• The KIPS Transactions:PartB
• /
• v.14B no.2
• /
• pp.81-88
• /
• 2007

We implement automobile license plates recognition system. These days automobile license plate recognition systems are widely used for tracing stolen cars. managing parking facilities, ticketing speeding cars, and so on. Recognition systems largely consist of three parts plates extraction, segments extraction, and segment recognition. For plates extraction, we measure the degree of inclination of plate. We use filters that extract only the horizontal components of the front of an automobile to measure the degree of inclination. For segment extraction, we trace the change of the number of blocks that consist solely of foreground pixels or background pixels as the horizontal scanning line moves along upward. For recognition of each individual letter or digit, we devise a variant of template matching method, called comparative template matching. Through experiments, we show that comparative template matching is less prone misled by noises and exhibits higher performance compared to the traditional method of template matching or histogram based recognition.