Proceedings of the KOSOMBE Conference (대한의용생체공학회:학술대회논문집)

The Korean Society of Medical and Biological Engineering (대한의용생체공학회)
권호 표지

The Study on the Medical Image Compression using the Characteristics of Human Visual System

인간 시각 장치의 특성을 이용한 의학 영상 압축에 관한 연구

  • Chee, Young-Joon (The Department of Biomedical Engineering, Seoul National University) ;
  • Park, Kwang-Seok (The Department of Biomedical Engineering, Seoul National University)
  • 지영준 (서울대학교 대학원 의용 생체 공학 협동 과정) ;
  • 박광석 (서울대학교 의과대학 의공학 교실)
  • Published : 1993.05.14
Download PDF
⟨ Previous Next ⟩

Abstract

For efficient transmission and storage of digital images, the requirements of image compression is incresing. Because the medical images contain diagnostic information small distortion has been more important factor than the low rate in such images. Generally the distortion in image is the difference of pixel values. However the image is percieved by human visual systems. So it is reasonable that human visual system characteristics be used as criteria of the image compression. In this paper, the Just Noticeable Difference curve is used as criteria of determining the homogeniety of a block and acceptibility of distortions. And Block Truncation Coding using spatial masking effect of eyes is adopted to code the blocks which contain line components. And small blocks which varies slowly can be approximated to polynomial functions successfully. We proposed the hybrid block coding scheme based on the block characteristics and human visual system characteristics. Simulation to several kinds of the medical images using this method showed that medical images can be compressed 5:1 - 10:1 without noticeable distortion.

Keywords