• The KIPS Transactions:PartB
• /
• v.12B no.6 s.102
• /
• pp.703-714
• /
• 2005

Researches in text categorization have been confined to whole-document-level classification, probably due to lacks of full-text test collections. However, full-length documents availably today in large quantities pose renewed interests in text classification. A document is usually written in an organized structure to present its main topic(s). This structure can be expressed as a sequence of sub-topic text blocks, or passages. In order to reflect the sub-topic structure of a document, we propose a new passage-level or passage-based text categorization model, which segments a test document into several Passages, assigns categories to each passage, and merges passage categories to document categories. Compared with traditional document-level categorization, two additional steps, passage splitting and category merging, are required in this model. By using four subsets of Routers text categorization test collection and a full-text test collection of which documents are varying from tens of kilobytes to hundreds, we evaluated the proposed model, especially the effectiveness of various passage types and the importance of passage location in category merging. Our results show simple windows are best for all test collections tested in these experiments. We also found that passages have different degrees of contribution to main topic(s), depending on their location in the test document.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.24 no.1
• /
• pp.21-28
• /
• 2000

Although commercial PIV systems have been widely used for the non-intrusive velocity field measurement of fluid flows, they are still under development and have considerable room for improvement. In this study, a single-frame double-exposure PIV system using a high-resolution CCD camera was developed. A pulsed Nd:Yag laser and high-resolution CCD camera were synchronized by a home-made control circuit. In order to resolve the directional ambiguity problem encountered in the single-frame PIV technique, the second particle image was genuinely shifted in the CCD sensor array during the time interval dt. The velocity vector field was determined by calculating the displacement vector at each interrogation window using cross-correlation with 50% overlapping. In order to check the effect of spatial resolution of CCD camera on the accuracy of PIV velocity field measurement, the developed PIV system with three different resolution modes of the CCD camera (512 ${\times}$ 512, lK ${\times}$ IK, 2K ${\times}$ 2K) was applied to a turbulent flow which simulate the Zn plating process of a steel strip. The experimental model consists of a snout and a moving belt. Aluminum flakes about $1{\mu}m$ diameter were used as scattering particles for the liquid flow in the zinc pot and the gas flow above the zinc surface was seeded with atomized olive oil with an average diameter of 1-$3{\mu}m$. Velocity field measurements were carried out at the strip speed $V_s$=1.0 m/s. The 2K ${\times}$ 2K high-resolution PIV technique was significantly superior compared to the smaller pixel resolution PIV system. For the cases of 512 ${\times}$ 512 and 1K ${\times}$ 1K pixel resolution PIV system, it was difficult to get accurate flow structure of viscous flow near the wall and small vortex structure in the region of large velocity gradient.

• The Journal of the Acoustical Society of Korea
• /
• v.30 no.6
• /
• pp.303-307
• /
• 2011

This paper describes an audio watermarking using Quantization Index Modulation (QIM) on significant peaks in frequency domain. The audio signal is broken up into L samples length frames with non-overlapping and rectangular window. The zero-crossing rate of each frame is calculated for decision whether it is proper to be watermarked or not. If the frame is legitimate, frequency magnitude response is computed by discrete Fourier transform. For the QIM, we set the quantization step size based on maximum value of frequency magnitude response and select n significant peaks with w samples around them in frequency domain, totally $n{\times}(w+1)$ samples. Finally, watermark embedding is performed. Decoder extract watermarks based on Euclidean distance, that is a blind detection. The proposed method is robust against many attacks of watermark benchmark.