Abstract
In this study, we proposed the distances of PTP sheet designs to measure their similarities. Under the assumption that the design repeats in the direction of row, it was decomposed into the parts of repetition unit. We employed template matching to obtain the best matching of the parts, and defined intermediate expressions of the designs to absorb the aperiodicity in the designs of the parts. We defined two distances, the extension of Euclid distance of the intermediate expressions in CIE L*a*b* combining low pass filter in frequency domain, and the distance to measure the difference of the shifts. We calculated the distances for each pair of 198 PTP designs, and searched the designs that have short distance to randomly selected input designs. As a result, for three input images, we could extract PTP sheet designs similar to the inputs. It was also found that, if the ratio of forecolor part to background color part is small, the distance tends to have small value. Because of this, it is necessary to limit the use of dint to search PTP sheet images which is similar to an input image.
Original language | English |
---|---|
Pages (from-to) | 208-215 |
Number of pages | 8 |
Journal | Procedia Manufacturing |
Volume | 3 |
DOIs | |
Publication status | Published - 2015 Jan 1 |
Keywords
- Image processing
- Medical safety
- PTP sheet design
- Similarity
ASJC Scopus subject areas
- Industrial and Manufacturing Engineering
- Artificial Intelligence