Evaluation of the effective information preservation method for binary image reduction

Abstract

AbstractOne effective technique for searching through a huge volume of stored binary documents and graphics is to use high‐quality, high‐compression reduction so that a large quantity of pictures can be presented at one time for browsing. In this connection, the author has previously reported on the usefulness of the Extended Thin Line Preservation (ETP) methods in binary image reduction. This paper proposes an enhancement of ETP called the Effective Information Preservation (EIP) method that achieves even higher compression and quality by preserving effective information existing in the original image. The paper begins by describing the principle behind preservation of effective information in the EIP method and showing that a converted picture without missing information can be generated. It then describes subjective evaluation experiments that were performed to determine the effectiveness of the EIP conversion method in terms of picture quality, using the Binary to Gray‐Scale (BGS) method for converting to halftone pictures as a quality index. Results of the experiments revealed that the EIP method produces better pictures than the BGS method in hardcopy evaluation, and that quality was favorable for conversion ratios under 1/5 in softcopy evaluation. Furthermore, in comparing the two methods for the same amount of data, the EIP method was found to exhibit extremely high data efficiency. These findings indicate that EIP can produce high‐quality pictures with a small amount of data for high‐compression reduction at a ratio of 1/10, and that it is useful in creating a superlarge multi‐image display for high‐speed browsing. © 2001 Scripta Technica, Syst Comp Jpn, 32(7): 1–11, 2001