ABSTRACT
Halftone images produced by error diffusion algorithms suffer from texture structures that decrease their visual qualities, such as worm-like and stripe-like defects. An effective quality evaluation metric is needed to evaluate various error diffusion halftoning algorithms and their corresponding halftone images. A texture distortion evaluation metric (TDEM) is proposed to measure the texture distortions within the referenced halftoned greyscale images. First, the weights of different tonal regions were assigned based on the two types of visual objects available and the locations of the texture structures. Then, based on image structure correlations and human visual characteristics, a block operation was performed and TDEM values were obtained by summing the local block-weighted variances based on image information. The proposed metric was validated using objective comparison and psycho-visual experiments. Results indicate that the metric can measure texture distortions effectively, and that the results are consistent with subjective visual perception.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes on contributors
Yaohua Yi received B.S.E. and M.S. from the School of Printing and Packaging from Wuhan University in 1997 and 2000, respectively. He received Ph.D. from the State Key Laboratory for Information Engineering in Surveying, Mapping and Remote sensing, Wuhan University, Wuhan, China, in 2004. He has been a professor in the School of Printing and Packaging, Wuhan University, since 2010. His current interest includes digital imaging processing and image colour management.
Rui Li received B.S.E. from Nanjing Forestry University, Nanjing, China, in 2015. She is pursuing her M.S. in the School of Printing and Packaging, Wuhan University. Her current research interests include image quality assessment and halftoning algorithm evaluation.
Changhui Yu received B.S.E. from Henan University, Kaifeng, Henan, China, in 1997, M.S. from Wuhan University, Wuhan, China, in 2000, and Ph.D. from the School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, China, in 2007. She has been an associate professor in the School of Remote Sensing and Information Engineering, Wuhan University, since 2009. Her research interests include remote sensing image processing, geographic information system, big data analysis, data provenance and image quality assessment.
Yuan Yuan received B.S.E. and M.S. in computer science and technology from Wuhan University, Wuhan, China, in 2011 and 2013, respectively. She is doing her PhD at the School of Printing and Packaging, Wuhan University. Her current research interests include image quality assessment, deep learning and image processing.