A bit‐plane decomposition method of multivalued images has the advantage in an interactive image retrieving system whereby users obtain a quick overview of an entire picture and a detailed local configuration simultaneously. The simplicity and feasibility of this method have attracted many researchers to produce numerous works in this field. Yet no one has been quite successful in reducing data amounts very effectively. The reason is that they tried to preserve all image information without paying much attention to the relative importance of each local information to the entire image. This paper proposes a data reduction strategy which incorporates a part of the property of human vision systems. Fine and exact data of a region where discontinuous points are very dense are neglected and abolished. On the other hand, all information about a region where discontinuous points are sparse is preserved. First, the properties of black‐and‐white (B/W) boundary points on image bit‐planes are studied and a new data compression method is presented. The highlight of this method is the computation of a measure for the density of the discontinuous points. Then an alternative measure is shown which is based on an image complexity of B/W pictures, and is less complicated to compute. It provides a data compression strategy which can be followed on coded data of the image. The experimental study supported this new idea, namely, most of the edge parts were enhanced and nonedge areas were noise‐cleaned, and a good data compression was obtained.
ASJC Scopus subject areas
- Theoretical Computer Science
- Information Systems
- Hardware and Architecture
- Computational Theory and Mathematics