Title of article :
Effective Digital Image Colors Reduction/Quantization Method
Author/Authors :
Mohamme, Faisel G. University of Baghdad - College of Science - Department of Computer Science, Iraq , Amer, Wejdan A. University of Baghdad - College of Science - Department of Computer Science, Iraq
Abstract :
In the current research work, a method to reduce the color levels of the pixels within digital images was proposed. The recent strategy was based on self organization map neural network method (SOM). The efficiency of recent method was compared with the well known logarithmic methods like Floyd-Steinberg (Halftone) dithering and Octtrees (Quadtrees) methods. Experimental results have shown that by adjusting the sampling factor can produce higher-quality images with no much longer run times, or some better quality with shorter running times than existing methods. This observation refutes the repeated neural networks is necessarily slow but have best results. The generated quantization map can be exploited for color image compression, classification, or to edit the color palette for different image graphical applications.
Keywords :
Quantization , SOM neural network algorithm , Color reduction , Classification
Journal title :
Iraqi Journal Of Science
Journal title :
Iraqi Journal Of Science