最近项目调研,要是用百度智能云的ocr文字识别功能,虽然说官方文档和demo都很详细,但是在实际操作的过程中,还是遇到很多问题,那就记录一下吧~
1)注册
首先要在百度智能云上进行注册,注册后进行应用创建。
注册后效果:
2)创建应用
我个人创建的应用如下:
3)获取Access Token
官方说明:
https://ai.baidu.com/ai-doc/REFERENCE/Ck3dwjhhu
4)demo编写
ocr识别官方提供两种方式,一种是SDK,一种是API接口(官方测试demo采用curl)
接口说明文档:
https://ai.baidu.com/ai-doc/OCR/jk9m7mj1l
主要原理:
读取图片内容------>转成Base64----->转成urlencode编码
定额发票识别代码示例:
bool MainWindow::syncPost()
{
QByteArray replyData;
QFile original_file("D:\\Test\\ocrTest\\1234567.jpeg");
if (!original_file.open(QIODevice::ReadOnly))
{
return false;
}
QByteArray encrypted_data = original_file.readAll().toBase64();
original_file.close();
QByteArray post_data;
post_data.append("image=");
post_data.append(encrypted_data.toPercentEncoding());
bool ret = false;
QNetworkAccessManager networkAccessManager;
QNetworkRequest request(QUrl("https://aip.baidubce.com/rest/2.0/ocr/v1/quota_invoice?access_token=xxxxx"));
request.setRawHeader("User-Agent", "Mozilla/5.0 (compatible; MSIE 10.0; Windows NT 6.2;Trident/6.0)");
request.setRawHeader("Connection", "close");
request.setRawHeader("Content-Type", "application/x-www-form-urlencoded");
request.setHeader(QNetworkRequest::ContentTypeHeader, "application/x-www-form-urlencoded");
QNetworkReply* reply = networkAccessManager.post(request, post_data.data());
QEventLoop loop;
QObject::connect(reply, &QNetworkReply::finished, &loop, &QEventLoop::quit);
loop.exec();
int statusCode = reply->attribute(QNetworkRequest::HttpStatusCodeAttribute).toInt();
if (reply->error() == QNetworkReply::NoError && statusCode == 200)
{
ret = true;
replyData = reply->readAll();
}
reply->deleteLater();
return ret;
}
注:1)有些功能Url是免费测试
2)官方使用curl测试,我的demo使用Qt的http请求