Edited at

Golangで画像処理をしたい

More than 5 years have passed since last update.

(処理自体は精度低い or 間違っているかも)

imageパッケージが便利

ちょっとハマったのは image/jpegをimportしないとjpegのDecodeが動かないということ(pngならimage/png)

Image structは.At(x, y)で指定したpixelの色データが取れるのが便利だ

package main

import (
"fmt"
"github.com/nfnt/resize"
"image"
_ "image/jpeg"
"os"
)

func main() {
file, err := os.Open("image.jpg")
defer file.Close()
if err != nil {
fmt.Println(err)
return
}
img, _, err := image.Decode(file)
if err != nil {
fmt.Println(err)
return
}

// resize for performance
img = resize.Resize(36, 0, img, resize.Lanczos3)

hist := getHistgram(img)

max := 0
// find most appered color
for _, val := range hist {
if val > max {
max = val
}
}
r, g, b := int2rgb(max)
colorCode := "#" + dec2hex(r, 2) + dec2hex(g, 2) + dec2hex(b, 2)

// most appered color in image
println(colorCode)
}

func getHistgram(img image.Image) []int {
hist := make([]int, 1000000)

// get bounds
rect := img.Bounds()
// color reduction
for i := 0; i < rect.Max.Y; i++ {
for j := 0; j < rect.Max.X; j++ {
r, g, b, _ := img.At(j, i).RGBA()
i := rgb2int(int(r), int(g), int(b))
hist[i]++
}
}
return hist
}

func dec2hex(n, beam int) string {
hex := ""
str := "0123456789abcdef"
for i := 0; i < beam; i++ {
m := n & 0xf
hex = string(str[m]) + hex
n -= m
n >>= 4
}
return hex
}

func rgb2int(r, g, b int) int {
return (((r >> 5) << 6) | ((g >> 5) << 3) | ((b >> 5) << 0))
}

func int2rgb(i int) (r, g, b int) {
return ((i >> 6 & 0x7) << 5) + 16, ((i >> 3 & 0x7) << 5) + 16, ((i >> 0 & 0x7) << 5) + 16
}