使用 Swift 进行验证码识别：集成 Tesseract OCR

1. 环境准备
   1.1 安装 Tesseract OCR
   更多内容访问ttocr.com或联系1436423940
   在 macOS 上可以使用 Homebrew 进行安装：

brew install tesseract

安装完成后，检查 Tesseract 是否安装成功：

tesseract --version

1.2 创建 Swift 项目

如果是 macOS 应用，可以使用 Swift Package Manager (SPM)，或者直接在 Xcode 项目中集成 OCR 识别功能。

创建一个新的 Swift 项目：

mkdir SwiftCaptchaOCR
cd SwiftCaptchaOCR
swift package init --type executable

编辑 Package.swift，添加 Tesseract 相关库：

// swift-tools-version:5.5
import PackageDescription

let package = Package(
name: "SwiftCaptchaOCR",
dependencies: [
.package(url: "https://github.com/gali8/Tesseract-OCR-iOS.git", from: "4.0.0")
],
targets: [
.executableTarget(
name: "SwiftCaptchaOCR",
dependencies: ["Tesseract-OCR-iOS"]
)
]
)

然后运行：

swift build

2. 代码实现

在 Sources/SwiftCaptchaOCR/main.swift 中写入以下代码：

import Foundation
import TesseractOCR
import Cocoa

func preprocessImage(inputPath: String, outputPath: String) {
guard let image = NSImage(contentsOfFile: inputPath) else {
print("无法加载图片")
return
}

// 转换为灰度图像
let grayscaleImage = convertToGrayscale(image: image)// 二值化处理
let binaryImage = applyThreshold(image: grayscaleImage, threshold: 128)// 保存处理后的图片
saveImage(image: binaryImage, outputPath: outputPath)

}

func convertToGrayscale(image: NSImage) -> NSImage {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)
let grayscaleRep = rep?.converting(to: .deviceGray, renderingIntent: .default)
let grayImage = NSImage(size: image.size)
grayImage.addRepresentation(grayscaleRep!)
return grayImage
}

func applyThreshold(image: NSImage, threshold: CGFloat) -> 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)!.whiteComponentlet newColor = color > threshold / 255.0 ? NSColor.white : NSColor.blackrep.setColor(newColor, atX: x, y: y)}
}let newImage = NSImage(size: image.size)
newImage.addRepresentation(rep)
return newImage

}

func saveImage(image: NSImage, outputPath: String) {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)
let pngData = rep?.representation(using: .png, properties: [:])
try? pngData?.write(to: URL(fileURLWithPath: outputPath))
}

func recognizeCaptcha(imagePath: String) -> String {
guard let tesseract = G8Tesseract(language: "eng") else {
return "Tesseract 初始化失败"
}
tesseract.image = NSImage(contentsOfFile: imagePath)
tesseract.recognize()
return tesseract.recognizedText ?? "识别失败"
}

let inputImagePath = "captcha.png" // 请替换为你的验证码图片路径
let processedImagePath = "processed_captcha.png"

// 预处理验证码图像
preprocessImage(inputPath: inputImagePath, outputPath: processedImagePath)

// OCR 识别
let result = recognizeCaptcha(imagePath: processedImagePath)
print("识别出的验证码: (result)")

3. 代码解析
   3.1 图像预处理

为了提高 OCR 识别率，我们进行了以下优化：

转换为灰度图像：

func convertToGrayscale(image: NSImage) -> NSImage {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)
let grayscaleRep = rep?.converting(to: .deviceGray, renderingIntent: .default)
let grayImage = NSImage(size: image.size)
grayImage.addRepresentation(grayscaleRep!)
return grayImage
}

二值化处理，增强字符对比度：

func applyThreshold(image: NSImage, threshold: CGFloat) -> NSImage {
let rep = NSBitmapImageRep(data: image.tiffRepresentation!)!
for x in 0..<rep.pixelsWide {
for y in 0..<rep.pixelsHigh {
let color = rep.colorAt(x: x, y: y)!.whiteComponent
let newColor = color > threshold / 255.0 ? NSColor.white : NSColor.black
rep.setColor(newColor, atX: x, y: y)
}
}
let newImage = NSImage(size: image.size)
newImage.addRepresentation(rep)
return newImage
}

3.2 OCR 解析

初始化 Tesseract OCR：

guard let tesseract = G8Tesseract(language: "eng") else {
return "Tesseract 初始化失败"
}

加载图像并执行 OCR：

tesseract.image = NSImage(contentsOfFile: imagePath)
tesseract.recognize()
tesseract.recognizedText ?? "识别失败"

4. 运行程序

确保 captcha.png 在项目目录下，然后运行：

swift run

示例输出：

识别出的验证码: X9F2G

5. 提高 OCR 识别率
   5.1 设置 Tesseract PSM 模式

Tesseract 提供不同的页面分割模式（PSM），可以调整以优化验证码识别：

tesseract.setVariableValue("6", forKey: "tessedit_pageseg_mode")

5.2 只识别特定字符
tesseract.charWhitelist = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"

5.3 进一步优化

如果验证码干扰较多，可以使用 Core Image 进行滤波、去噪等处理。