Golang의 이미지 분할 및 콘텐츠 인식 방법

go get github.com/otiai10/gosseract/v2
go get -u gonum.org/v1/gonum/...

package main

import (
    "fmt"
    "image"
    "image/color"
    "image/jpeg"
    "os"
)

func main() {
    file, err := os.Open("image.jpg")
    if err != nil {
        fmt.Println("图片读取失败：", err)
        return
    }
    defer file.Close()

    img, err := jpeg.Decode(file)
    if err != nil {
        fmt.Println("图片解码失败：", err)
        return
    }

    gray := image.NewGray(img.Bounds())
    for x := gray.Bounds().Min.X; x < gray.Bounds().Max.X; x++ {
        for y := gray.Bounds().Min.Y; y < gray.Bounds().Max.Y; y++ {
            r, g, b, _ := img.At(x, y).RGBA()
            grayColor := color.Gray{(r + g + b) / 3}
            gray.Set(x, y, grayColor)
        }
    }
}

package main

import (
    "fmt"
    "image"
    "image/color"
    "image/jpeg"
    "math"
    "os"
)

func main() {
    // ...

    // 分割图片
    bounds := gray.Bounds()
    threshold := otsu(gray) // OTSU算法获取阈值
    binary := image.NewGray(bounds)
    for x := bounds.Min.X; x < bounds.Max.X; x++ {
        for y := bounds.Min.Y; y < bounds.Max.Y; y++ {
            if gray.GrayAt(x, y).Y > threshold {
                binary.Set(x, y, color.Gray{255})
            } else {
                binary.Set(x, y, color.Gray{0})
            }
        }
    }
}

// OTSU算法计算阈值
func otsu(img *image.Gray) uint32 {
    var hist [256]int
    bounds := img.Bounds()
    for x := bounds.Min.X; x < bounds.Max.X; x++ {
        for y := bounds.Min.Y; y < bounds.Max.Y; y++ {
            hist[img.GrayAt(x, y).Y]++
        }
    }

    total := bounds.Max.X * bounds.Max.Y
    var sum float64
    for i := 0; i < 256; i++ {
        sum += float64(i) * float64(hist[i])
    }
    var sumB float64
    wB := 0
    wF := 0
    var varMax float64
    threshold := 0

    for t := 0; t < 256; t++ {
        wB += hist[t]
        if wB == 0 {
            continue
        }
        wF = total - wB
        if wF == 0 {
            break
        }
        sumB += float64(t) * float64(hist[t])

        mB := sumB / float64(wB)
        mF := (sum - sumB) / float64(wF)

        var between float64 = float64(wB) * float64(wF) * (mB - mF) * (mB - mF)
        if between >= varMax {
            threshold = t
            varMax = between
        }
    }

    return uint32(threshold)
}

package main

import (
    "fmt"
    "image"
    "image/color"
    "image/jpeg"
    "os"
    "strings"

    "github.com/otiai10/gosseract/v2"
)

func main() {
    // ...

    client := gosseract.NewClient()
    defer client.Close()

    texts := make([]string, 0)
    bounds := binary.Bounds()
    for x := bounds.Min.X; x < bounds.Max.X; x++ {
        for y := bounds.Min.Y; y < bounds.Max.Y; y++ {
            if binary.GrayAt(x, y).Y == 255 {
                continue
            }
            sx := x
            sy := y
            ex := x
            ey := y
            for ; ex < bounds.Max.X && binary.GrayAt(ex, y).Y == 0; ex++ {
            }
            for ; ey < bounds.Max.Y && binary.GrayAt(x, ey).Y == 0; ey++ {
            }
            rect := image.Rect(sx, sy, ex, ey)
            subImg := binary.SubImage(rect)

            pix := subImg.Bounds().Max.X * subImg.Bounds().Max.Y
            blackNum := 0
            for i := subImg.Bounds().Min.X; i < subImg.Bounds().Max.X; i++ {
                for j := subImg.Bounds().Min.Y; j < subImg.Bounds().Max.Y; j++ {
                    if subImg.At(i, j) == color.Gray{255} {
                        blackNum++
                    }
                }
            }
            if float64(blackNum)/float64(pix) < 0.1 { // 去除噪音
                continue
            }

            output, _ := client.ImageToText(subImg)
            output = strings.ReplaceAll(output, "
", "")
            output = strings.ReplaceAll(output, " ", "")
            texts = append(texts, output)
        }
    }

    fmt.Println(texts)
}