private void ChangetoSearchablePdf(string[] folders)
{
string incomefolder = folders[0];
string outcomefolder = folders[1];
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
string[] files = System.IO.Directory.GetFiles(incomefolder, "*.pdf").Select(Path.GetFileName).ToArray();
foreach (string tmpfile in files)
{
try {
string fullinputfilename = incomefolder + @"/" + tmpfile;
string fulloutputfilename = outcomefolder + @"/" + tmpfile;
ocr.Recognize(fullinputfilename, -1, -1, -1, -1, -1,
AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PDF,
AspriseOCR.PROP_PDF_OUTPUT_FILE, fulloutputfilename,
AspriseOCR.PROP_PDF_OUTPUT_TEXT_VISIBLE, false);
Console.WriteLine("File " + fulloutputfilename + "has done transformation yeah");
File.Delete(fullinputfilename);
} catch (Exception ex) {
Console.WriteLine(ex + "3333");
}
}
ocr.StopEngine();
}
private void button2_Click(object sender, EventArgs e)
{
openFileDialog1.InitialDirectory = Path.Combine(Application.StartupPath, "test");
if (openFileDialog1.ShowDialog() != DialogResult.OK)
{
return;
}
//PictureBox控件显示图片
pictureBox1.Load(openFileDialog1.FileName);
//获取用户选择文件的后缀名
string extension = Path.GetExtension(openFileDialog1.FileName);
//声明允许的后缀名
string[] str = new string[] { ".jpg", ".png" };
if (!str.Contains(extension))
{
MessageBox.Show("仅能上传jpg,png格式的图片！");
}
else
{
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
string s = ocr.Recognize(openFileDialog1.FileName, -1, -1, -1, -1, -1, AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PLAINTEXT);
label1.Text = s;
ocr.StopEngine();
}
}
public static void GetOCRpart(string img_path)
{
//if (threadOcr != null && threadOcr.IsAlive)
//{
// return;
//}
requestLang = "eng";
if (requestLang == null || requestLang.Length == 0)
{
return;
}
requestPropsStart = "";
requestImgFile = img_path;
requestLayout = "auto";
requestDataCapture = true;
requestAutoRotate = false;
requestWordLevel = false;
requestOutputFormat = AspriseOCR.OUTPUT_FORMAT_PLAINTEXT;
requestPdfHighlight = true;
requestRecognizeType = AspriseOCR.RECOGNIZE_TYPE_ALL;
requestPropsRecognize = "";
AspriseOCR.SetUp();
ocr = new AspriseOCR();
ocr.StartEngine(currentLang, AspriseOCR.SPEED_FASTEST, "");
doOcr();
}
public AspOCR()
{
AspriseOCR.SetUp();
ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
IDictionary <String, Boolean> dict = new Dictionary <String, Boolean>();
dict.Add(AspriseOCR.PROP_OUTPUT_SEPARATE_WORDS, true);
}
public static string ReadImage()
{
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
string s = ocr.Recognize("D:\\DouYin\\1539354248783.jpg", -1, -1, -1, -1, -1, AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PLAINTEXT);
ocr.StopEngine();
return(s);
}
private void btnOcrRead_Click(object sender, EventArgs e)
{
//textResults.Text = Marshal.PtrToStringAnsi(OCR(textImage.Text, -1));
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
string s = ocr.Recognize(textImage.Text, -1, -1, -1, -1, -1,
AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PLAINTEXT);
textResults.Text = s;
// process more images here ...
ocr.StopEngine();
}
private void button1_Click(object sender, EventArgs e)
{
OpenFileDialog openFileDialog = new OpenFileDialog();
if (openFileDialog.ShowDialog() == DialogResult.OK)
{
//PictureBox控件显示图片
pictureBox1.Load(openFileDialog.FileName);
//获取用户选择文件的后缀名
string extension = Path.GetExtension(openFileDialog.FileName);
//声明允许的后缀名
string[] str = new string[] { ".jpg", ".png" };
if (!str.Contains(extension))
{
MessageBox.Show("仅能上传jpg,png格式的图片！");
}
else
{
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("chi_sim", AspriseOCR.SPEED_FASTEST);
string s = ocr.Recognize(openFileDialog.FileName, -1, -1, -1, -1, -1,
AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PLAINTEXT);
//Console.WriteLine("OCR Result: " + s);
richTextBox1.Text = s;
// process more images here ...
ocr.StopEngine();
//识别图片文字
//var img = new Bitmap(openFileDialog.FileName);
//var ocr = new TesseractEngine(@"H:\Tesseract-OCR\tessdata\tesseract-ocr-3.02.chi_sim\tesseract-ocr\tessdata", "chi_sim", EngineMode.TesseractAndCube);
//var page = ocr.Process(img);
//richTextBox1.Text = page.GetText();
}
}
}
public string imgdo(Bitmap img)
{
Bitmap btp = img;
#region 前期处理
int Dggay = GetDgGrayValue(btp);
btp = ClearNoise(btp, Dggay, 1);
btp = ClearNoise(btp, Dggay, 3);
btp = ClearNoise(btp, Dggay);
btp.Save("d:\\去除相关颜色.png");
#endregion
//去色
Color c = new Color();
int rr, gg, bb;
for (int i = 0; i < btp.Width; i++)
{
for (int j = 0; j < btp.Height; j++)
{
//取图片当前的像素点
c = btp.GetPixel(i, j);
rr = c.R; gg = c.G; bb = c.B;
//改变颜色
if (rr == 102 && gg == 0 && bb == 0)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
if (rr == 153 && gg == 0 && bb == 0)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
if (rr == 153 && gg == 0 && bb == 51)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
if (rr == 153 && gg == 43 && bb == 51)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
if (rr == 255 && gg == 255 && bb == 0)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
if (rr == 255 && gg == 255 && bb == 51)
{
//重新设置当前的像素点
btp.SetPixel(i, j, Color.FromArgb(255, 255, 255, 255));
}
}
}
// btp.Save("d:\\去除相关颜色.png");
// pictureBox2.Image = Image.FromFile("d:\\去除相关颜色.png");
//灰度
Bitmap bmphd = btp;
for (int i = 0; i < bmphd.Width; i++)
{
for (int j = 0; j < bmphd.Height; j++)
{
//取图片当前的像素点
var color = bmphd.GetPixel(i, j);
var gray = (int)(color.R * 0.001 + color.G * 0.700 + color.B * 0.250);
//重新设置当前的像素点
bmphd.SetPixel(i, j, Color.FromArgb(gray, gray, gray));
}
}
// bmphd.Save("d:\\灰度.png");
//pictureBox27.Image = Image.FromFile("d:\\灰度.png");
//二值化
Bitmap erzhi = bmphd;
Bitmap orcbmp;
int nn = 3;
int w = erzhi.Width;
int h = erzhi.Height;
BitmapData data = erzhi.LockBits(new Rectangle(0, 0, w, h), ImageLockMode.ReadOnly, PixelFormat.Format24bppRgb);
unsafe
{
byte *p = (byte *)data.Scan0;
byte[,] vSource = new byte[w, h];
int offset = data.Stride - w * nn;
for (int y = 0; y < h; y++)
{
for (int x = 0; x < w; x++)
{
vSource[x, y] = (byte)(((int)p[0] + (int)p[1] + (int)p[2]) / 3);
p += nn;
}
p += offset;
}
erzhi.UnlockBits(data);
Bitmap bmpDest = new Bitmap(w, h, PixelFormat.Format24bppRgb);
BitmapData dataDest = bmpDest.LockBits(new Rectangle(0, 0, w, h), ImageLockMode.WriteOnly, PixelFormat.Format24bppRgb);
p = (byte *)dataDest.Scan0;
offset = dataDest.Stride - w * nn;
for (int y = 0; y < h; y++)
{
for (int x = 0; x < w; x++)
{
p[0] = p[1] = p[2] = (int)vSource[x, y] > 161 ? (byte)255 : (byte)0;
//p[0] = p[1] = p[2] = (int)GetAverageColor(vSource, x, y, w, h) > 50 ? (byte)255 : (byte)0;
p += nn;
}
p += offset;
}
bmpDest.UnlockBits(dataDest);
orcbmp = bmpDest;
orcbmp.Save(Environment.CurrentDirectory + "\\二值化.png");
// pictureBox29.Image = Image.FromFile("d:\\二值化.png");
}
//OCR的值
if (orcbmp != null)
{
AspriseOCR.SetUp();
AspriseOCR ocr = new AspriseOCR();
ocr.StartEngine("eng", AspriseOCR.SPEED_FASTEST);
string s = ocr.Recognize(Environment.CurrentDirectory + "\\二值化.png", -1, -1, -1, -1, -1, AspriseOCR.RECOGNIZE_TYPE_ALL, AspriseOCR.OUTPUT_FORMAT_PLAINTEXT);
ocr.StopEngine();
return(s);
//string result = Ocr(orcbmp);
//label32.Text = result.Replace("\n", "\r\n").Replace(" ", "");
}
return("");
}
private void asprise_init()
{
bool browsed = false;
int count = 0;
string dllFilePath = AspriseOCR.getOcrDllPath();
if (dllFilePath == null)
{
string parentFolder = AspriseOCR.detectOcrDllInParentFolders();
if (parentFolder != null)
{
AspriseOCR.addToSystemPath(parentFolder);
log("Folder containing ocr dll detected: " + parentFolder);
}
}
String lang = GetLastLang();
if (lang == null || lang.Length == 0)
{
comboLang.SelectedIndex = 0;
currentLang = "eng";
}
else
{
for (int i = 0; i < comboLang.Items.Count; i++)
{
if (comboLang.Items[i].ToString().Equals(lang))
{
comboLang.SelectedIndex = i;
break;
}
}
currentLang = lang;
}
// Let user browse the ocr dll if it is not found in PATH.
while (true)
{
dllFilePath = AspriseOCR.getOcrDllPath();
if (dllFilePath == null)
{
log("OCR dll (" + AspriseOCR.getOcrDllName() + ") is not found in PATH.");
OpenFileDialog fileDialog = new OpenFileDialog();
fileDialog.RestoreDirectory = true;
fileDialog.Title = "Please select " + AspriseOCR.getOcrDllName();
fileDialog.FileName = AspriseOCR.getOcrDllName();
if (fileDialog.ShowDialog(this) == DialogResult.OK)
{
log("User browsed dll: " + fileDialog.FileName);
AspriseOCR.addToSystemPath(Path.GetDirectoryName(fileDialog.FileName));
browsed = true;
}
count++;
if (count == 2)
{
showMessageBox("OCR DLL not found. ", "Error");
break;
}
}
else
{
log("OCR dll found: " + dllFilePath);
if (browsed)
{
log("Please consider copy " + AspriseOCR.getOcrDllName() + " to directory: " +
Directory.GetCurrentDirectory());
}
log(AspriseOCR.GetLibraryVersion());
try
{
log("Starting OCR engine ...");
AspriseOCR.SetUp();
ocr = new AspriseOCR();
ocr.StartEngine(currentLang, AspriseOCR.SPEED_FASTEST);
log("OCR engine started successfully.");
}
catch (Exception e)
{
log("ERROR: Failed to start OCR engine: " + e);
log(e.StackTrace);
}
break;
}
}
// user preference
textImage.Text = GetLastImagePath();
}