Improving mid-tone quality of variable-coefficient error diffusion using threshold modulation

Zhou, Bingfeng; Fang, Xifeng

doi:10.1145/1201775.882289

Cited by 28 publications

(35 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As we can see from Table 1, our method consistently obtains the highest MSSIMs among the result sets. To be specific, our method is far better than the improving midtone E-D [8]. Besides, for the weak texture and dark tone region, the MSSIMs of our method are around 15 percent higher than those of Pang's method [11] and 30 percent higher than those of Chang's method [12] on average.…”

Section: Objective Quality Evaluationmentioning

confidence: 82%

“…Therefore, we test our algorithm with the above values in the following parts. Figures 3 and 4 visually compare our results to that of the improving mid-tone E-D [8], structure-aware halftoning [11] and structure-aware E-D [12]. Compared to the improving mid-tone E-D [8], our results better reproduce the range of contrasts and texture details undeniably.…”

Section: Feedback Coefficient Analysismentioning

confidence: 83%

“…Figures 3 and 4 visually compare our results to that of the improving mid-tone E-D [8], structure-aware halftoning [11] and structure-aware E-D [12]. Compared to the improving mid-tone E-D [8], our results better reproduce the range of contrasts and texture details undeniably. Besides, visual quality of our method is comparable to structure-aware halftoning which obtains the best visual effect among digital halftoning algorithms to date.…”

Section: Feedback Coefficient Analysismentioning

confidence: 83%

“…In order to reduce artificial textures in the E-D process, we adopt threshold modulation method of Zhou and Fang [8].…”

Section: Adaptive Median Filter and Threshold Modulationmentioning

confidence: 99%

“…However, due to generating worms and other undesirable artifacts such as noticeable directional textures, images produced by Floyd-Steinberg E-D [5] are obviously inferior to the itera-tive algorithms. Many methods were proposed to improve Floyd-Steinberg E-D. For instance, Zhou and Fang [8] introduced intensity-dependent noise to remove regular structures in critical intensity levels. Howerver, blurred texture details and poor contrast are two big drawbacks for this method.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

An improved error diffusion algorithm based on visual difference

Shi

2014

2014 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Based on the human vision system(HVS), this paper introduces an error diffusion algorithm that produces higher visual quality results while preserving fine texture details and tone consistency present in the original images. We calculate the visual difference between the original and halftone images based on the HVS model. Then by compensating original images for visual difference, reconstructed continuous-tone images that respect the human vision characteristics are obtained. Besides, we use the adaptive median filter to remove noises and preserve sharpness in the reconstructed images. Finally, Floyd-Steinberg E-D using threshold modulation is adopted to get the halftoning results. Multiple experiments show that our method outperforms the existing state-of-art halftoning algorithms and presents visually pleasing results with both fine texture details and clear lines especially in the weak texture and dark tone regions. Meanwhile, our method is much faster due to its simple processing procedure. Combined good visual properties with fast speed, our method can be applicable in most real-world applications.

show abstract

Section: Objective Quality Evaluationmentioning

confidence: 82%