Thanks to Julio Marchi for this space in MSX All
A.I. project aims at researching artifitial intelligence methods in order to help on identifying MSX colors.
2004 - The first related project uses Fuzzy Logic to identify MSX 1 colors palette.
2009 - MSX Viewer 5 (beta) is released bringing the Fuzzy Logic color identification and K-Means clustering for palette optimization. New fuzzy sets and rules are used as experimental work.
2017 - MSX Viewer 5 (version 5.0.1) replaces its fuzzy sets and rules by A.I. sets and rules.
1. Fuzzy Inference System
For a given PC input image, find the best color representation for that 24-bit image using only the 16 MSX 1 native colors. This approach do not make use of MSX 2 palette.
For a given PC input image, find the best MSX 2 or V9990 palette colors combination for that 24-bit color image.
The first experiments were based on the idea of using a Fuzzy Inference System to classify a PC format image according to the MSX 1 native palette.
- Windows binary program.
- It is able to ideintify MSX 1 colors using Fuzzy Logic.
- Size: 155 Kb.
- Paste an 24-bit color image by clicking on "Colar" button.
- Press "Fuzzy" button. The program identify the colors.
- Copy to memory, by pressing "Copiar". Then, paste on MSX Viewer 4.
- The image is ready to be saved as screen 2, 3, 4 or 7.
MSX Viewer 5 has the Fuzzy Logic at "Color Recognition" step. Thus, it is only able to recognize the MSX 1 default colors.
1. Reconhecimento de cores do MSX por Lógica Fuzzy
2. Otimização da Paleta de Cores
MSX 1 color adaptation using Fuzzy:
Using K-Means to optimize MSX 2 palette:
Note: for all examples above, the resulting images have only 16 colors.
Marcelo Teixeira Silveira
|© MarMSX 1999-2020|