extract_colors.md

Extracting colors from an image

To extract colors from an image that are suitable for generating schemes, use the quantize and score libraries.

TODO: Insert here a diagram for image-to-colors extraction.

Step 1 — Image to Pixels

First, convert an image into an array of pixels in ARGB format. MCU does not provide this feature; please use the idiomatic method provided by your language.

If the image is larger than 128 × 128 pixels, please scale the image down to 128 × 128 before converting it to an array of pixels. This ensures the speed of processing.

For example, in Java, one may use the BufferedImage.getRGB method:

import java.awt.image.BufferedImage;

class ImageUtils {
  // ...

  public static int[] imageToPixels(BufferedImage image) {
    int width = image.getWidth();
    int height = image.getHeight();
    BufferedImage outputImage = new BufferedImage(width, height, BufferedImage.TYPE_INT_ARGB);
    return image.getRGB(0, 0, width, height, null, 0, width);
  }
}

Step 2 — Pixels to Prominent Colors

Once you have the array of pixels, pass it into QuantizerCelebi.quantize provided by the quantize library.

Dart

final quantizerResult = await QuantizerCelebi.quantize(pixels, maxColors);

Java

QuantizerResult quantizerResult = QuantizerCelebi.quantize(pixels, maxColors);

TypeScript

const quantizerResult = QuantizerCelebi.quantize(pixels, maxColors);

C++

QuantizerResult quantizer_result = QuantizeCelebi(pixels, max_colors);

Swift

let quantizerResult = QuantizerCelebi().quantize(pixels, maxColors)

The parameter maxColors is a limit on the number of colors returned by the quantizer. A reasonable default is 128.

Step 3 — Prominent Colors to Suitable Seed Colors

Use the Score.score method provided by the score library to extract colors that are suitable as seeds for color schemes, ranked by decreasing suitability.

Dart

final colors = Score.score(quantizerResult.colorToCount);

Java

List<Integer> colors = Score.score(quantizerResult);

TypeScript

final colors = Score.score(quantizerResult);

C++

std::vector<Argb> colors = RankedSuggestions(quantizer_result.color_to_count);

Swift

let colors = Score.score(quantizerResult.colorToCount)