用 Swift 和 Tesseract OCR 识别验证码

1. 环境准备
   1.1 安装 Swift

macOS 默认自带 Swift，可用以下命令检查版本：

swift --version
更多内容访问ttocr.com或联系1436423940

如未安装，可从 Swift 官方网站
下载。

1.2 安装 Tesseract OCR

使用 Homebrew 安装：

brew install tesseract

安装后检查是否可用：

tesseract --version

1.3 创建 Swift 项目
mkdir SwiftOCR
cd SwiftOCR
swift package init --type executable

在 Package.swift 添加 TesseractOCR 依赖：

// swift-tools-version:5.5
import PackageDescription

let package = Package(
name: "SwiftOCR",
dependencies: [
.package(url: "https://github.com/gali8/Tesseract-OCR-iOS.git", from: "5.0.0")
],
targets: [
.executableTarget(
name: "SwiftOCR",
dependencies: ["TesseractOCR"])
]
)

2. 代码实现
   2.1 读取 & 处理图像

创建 Sources/SwiftOCR/main.swift，编写以下代码：

import Foundation
import TesseractOCR
import AppKit

func preprocessImage(imagePath: String) -> NSImage? {
guard let image = NSImage(contentsOfFile: imagePath) else {
print("无法加载图片")
return nil
}

// 转换为灰度图
let grayscaleImage = convertToGrayscale(image: image)

// 二值化处理
let binaryImage = binarizeImage(image: grayscaleImage)

return binaryImage

}

func convertToGrayscale(image: NSImage) -> NSImage {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)
let width = rep!.pixelsWide
let height = rep!.pixelsHigh

for x in 0..<width {
    for y in 0..<height {
        let color = rep!.colorAt(x: x, y: y)!
        let gray = (color.redComponent + color.greenComponent + color.blueComponent) / 3
        let newColor = NSColor(calibratedWhite: gray, alpha: 1)
        rep!.setColor(newColor, atX: x, y: y)
    }
}

let newImage = NSImage(size: image.size)
newImage.addRepresentation(rep!)
return newImage

}

func binarizeImage(image: NSImage) -> NSImage {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)
let width = rep!.pixelsWide
let height = rep!.pixelsHigh
let threshold: CGFloat = 0.5

for x in 0..<width {
    for y in 0..<height {
        let color = rep!.colorAt(x: x, y: y)!
        let gray = (color.whiteComponent > threshold) ? 1.0 : 0.0
        let newColor = NSColor(calibratedWhite: gray, alpha: 1)
        rep!.setColor(newColor, atX: x, y: y)
    }
}

let newImage = NSImage(size: image.size)
newImage.addRepresentation(rep!)
return newImage

}

func recognizeCaptcha(imagePath: String) {
guard let processedImage = preprocessImage(imagePath: imagePath) else {
return
}

let tesseract = G8Tesseract(language: "eng")
tesseract?.image = processedImage
tesseract?.recognize()

print("识别出的验证码: \(tesseract?.recognizedText ?? "识别失败")")

}

let imagePath = "captcha.png"
recognizeCaptcha(imagePath: imagePath)

3. 代码解析
   3.1 图像预处理

灰度化：

let gray = (color.redComponent + color.greenComponent + color.blueComponent) / 3

二值化：

let gray = (color.whiteComponent > threshold) ? 1.0 : 0.0

3.2 OCR 解析

初始化 Tesseract：

let tesseract = G8Tesseract(language: "eng")

设置图像并识别：

tesseract?.image = processedImage
tesseract?.recognize()

4. 运行程序

确保 captcha.png 存在于项目目录下，然后编译并运行：

swift run

示例输出：

识别出的验证码: 7XQ9T

5. 提高 OCR 识别准确率
   5.1 设定 Tesseract PSM 模式
   tesseract?.setVariableValue("6", forKey: "tessedit_pageseg_mode")

5.2 设定字符白名单
tesseract?.setVariableValue("0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ", forKey: "tessedit_char_whitelist")

5.3 进一步图像优化

去噪点

字符分割

6. 其他 OCR 方案

如果 Tesseract 识别率不够，可以使用 Vision 框架：

import Vision

func recognizeWithVision(image: NSImage) {
let request = VNRecognizeTextRequest { request, error in
guard let observations = request.results as? [VNRecognizedTextObservation] else { return }
for observation in observations {
if let topCandidate = observation.topCandidates(1).first {
print("识别出的验证码: (topCandidate.string)")
}
}
}

let handler = VNImageRequestHandler(cgImage: image.cgImage(forProposedRect: nil, context: nil, hints: nil)!, options: [:])
try? handler.perform([request])

}