Tesseract-ocr 3.0.2源码 + VS2010项目工程 + 简单测试代码

编译环境:Visual Studio 2010

所用类库版本:

zlib 1.2.7

lpng1514
jpegsr9
tiff-4.0.3
giflib-5.0.4
leptonica-1.69
tesseract-ocr3.0.2

下载地址http://pan.baidu.com/s/1c0pqvQg 密码rlib

测试代码

bool XYImage::RecognizeImage(const String &language /* = _T("eng") */,
							 const String &whitelist /* = _T("0123456789") */)
{
	tesseract::TessBaseAPI api;
	if (api.Init(GlobalizeString(AppBase::GetStartupPath()).toGBK(), GlobalizeString(language).toGBK(),
		tesseract::OEM_DEFAULT) == -1)
	{
		assert(!"ocr init failed!");
		return false;
	}

	api.SetVariable("classify_bln_numeric_mode", "1");
	api.SetVariable("tessedit_char_whitelist", GlobalizeString(whitelist).toGBK());

	auto pix = pixRead("F:\\1.bmp"); //把pixReadBmpStream函数从库里抄出来的, 可以实现加载内存图像
	if (pix != nullptr)
	{
		api.SetImage(pix);		

		char *pUTF8 = api.GetUTF8Text();//UTF-8编码

		if (pUTF8 != nullptr)
		{
			this->UTF8Text.Copy(pUTF8);
			delete[] pUTF8;
		}

		pixDestroy(&pix);

		return true;
	} //if

	return false;
}