Some tri-valued random patterns, for example tricolor of Red, Green and Blue, were generated using opposite-Sugeno integral filter. In the process of converting images, a conversion function for normalization was inserted between two steps of the filterings applied to independently Red, Green and Blue.
To make tri-valued patterns from multicolored patterns, they were firstly transformed into gray images. Secondly, the 3-tile thresholding was applied to segment the image into three regions. As the result, three kinds of binary images, and their three kinds of rotated binary images were obtained.
New type of symmetry was defined in addition reversal symmetry and rotational symmetry as defined in the previous paper. It was named as region reversal symmetry, was defined by equality in the index of contrast calculated from the first three kinds of binary images.
Ranking of tri-valued geometric patterns (14), random patterns (15) and collected camouflage patterns (135) was synthetically carried out in the evaluation of symmetry goodness using the three types of symmetry.
Ten ones out of the fifteen tri-valued random patterns were ranked in the twenty bests in the synthetic evaluation among all 164 patterns.
View full abstract