@article{ART003397389},
author={In-Kyoung Kim and Sung-woo Chon and Hyo-heon Kang},
title={A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={10},
pages={83-105}
TY - JOUR
AU - In-Kyoung Kim
AU - Sung-woo Chon
AU - Hyo-heon Kang
TI - A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 10
PB - The Korean Society Of Computer And Information
SP - 83
EP - 105
SN - 1598-849X
AB - This study defines and validates a distance-ridge-sampled thickness descriptor for AI-extracted animation line art. Twice the exact Euclidean distance transform (EDT) value was sampled and aggregated at the centers of maximal disks within the family of continuous open distance disks centered on foreground pixels. Nominal-width error and grid sensitivity were evaluated across 180 synthetic conditions and 3,234 conditions involving placement and geometry variations. Sensitivity to binarization, sampling, and aggregation was analyzed using 100 AniLines output images from a single nonpublic sequence. On the baseline grid, the mean absolute error was 1.712 pixels, higher than those of Skeleton/EDT and SWT. With equal weighting of ridge pixels, Otsu thresholding reduced mean thickness and the mean of per-image standard deviations by 11.3% and 39.9%, respectively, relative to a threshold of 1. However, the direction of change in mean thickness depended on sampling and weighting. The descriptor supports comparisons of binary structures under identical resolution, binarization, sampling, and aggregation conditions and is distinct from physical stroke width and perceptual quality scores.
KW - AI-Extracted Anime Line Art;Thickness Descriptor;Distance Ridge;Euclidean Distance Transform;Grid Discretization;Binarization Sensitivity
DO -
UR -
ER -
In-Kyoung Kim, Sung-woo Chon and Hyo-heon Kang. (2026). A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis. Journal of The Korea Society of Computer and Information, 31(10), 83-105.
In-Kyoung Kim, Sung-woo Chon and Hyo-heon Kang. 2026, "A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis", Journal of The Korea Society of Computer and Information, vol.31, no.10 pp.83-105.
In-Kyoung Kim, Sung-woo Chon, Hyo-heon Kang "A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis" Journal of The Korea Society of Computer and Information 31.10 pp.83-105 (2026) : 83.
In-Kyoung Kim, Sung-woo Chon, Hyo-heon Kang. A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis. 2026; 31(10), 83-105.
In-Kyoung Kim, Sung-woo Chon and Hyo-heon Kang. "A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis" Journal of The Korea Society of Computer and Information 31, no.10 (2026) : 83-105.
In-Kyoung Kim; Sung-woo Chon; Hyo-heon Kang. A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis. Journal of The Korea Society of Computer and Information, 31(10), 83-105.
In-Kyoung Kim; Sung-woo Chon; Hyo-heon Kang. A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis. Journal of The Korea Society of Computer and Information. 2026; 31(10) 83-105.
In-Kyoung Kim, Sung-woo Chon, Hyo-heon Kang. A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis. 2026; 31(10), 83-105.
In-Kyoung Kim, Sung-woo Chon and Hyo-heon Kang. "A Distance-Ridge Thickness Descriptor for AI-Extracted Anime Line Art: Definition, Validation, and Sensitivity Analysis" Journal of The Korea Society of Computer and Information 31, no.10 (2026) : 83-105.
