The Research of Replacing Swap with Toggle in Direct Binary Search Halftoning Algorithm

• Main Authors: Yu-Cheng Tsai, 蔡侑錚
• Other Authors: 廖俊睿
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/33629839214683096807

碩士 === 國立中興大學 === 電機工程學系所 === 105 === Direct binary search (DBS) is an image halftoning algorithm which can produce the best halftoned image quality. It simulates human visual system through a low-pass filter. It then uses a series of operations called toggles and swaps to obtain the lowest total squared perceived error. Although it can produce superior halftone quality, its computational load is high and its execution time is quite long. Through theoretical analysis, monotonically non-decreasing swap (MNDS) DBS separates the operations of toggle and swap. It first let the toggle operations converge and, then, proceed to the swap operations. However, the time required for swap to converge is much longer than toggle. In this thesis, we seek to use only the faster toggle operations without using swap operations. By using multiple rounds of toggle operations and using different filters in each round, we can produce an effect that is similar to swap. Considering the tradeoff between image quality and execution time, we find the best combinations of filters in the multiple rounds of toggles.