private void btnObjectAnalysis_Click(object sender, EventArgs e)
{
//use the PDF library to break out all of the PDF Objects
Readers.PDFReader Rdr = new Readers.PDFReader();
Rdr.getObjectsFromPDF(PDFFileName, out PDFPageWidth, out PDFPageHeight);
//break up text objects with multiple words into a single word per object
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.LocateKeywords();
PdfFixedDocument doc = new PdfFixedDocument(PDFFileName);
PdfPageRenderer renderer = new PdfPageRenderer(doc.Pages[0]);
FileStream PNGStream = File.OpenWrite(PDFFileName + ".png");
renderer.ConvertPageToImage(dpi, PNGStream, PdfPageImageFormat.Png);
PNGStream.Flush();
PNGStream.Close();
LoadImageBox();
PopulateGridView("Text");
}
private void AutoprocessDocs(string Filename)
{
//load the imagebox
PDFFileName = Filename;
//Autoprocess
//use the PDF library to break out all of the PDF Objects
Readers.PDFReader Rdr = new Readers.PDFReader();
Rdr.getObjectsFromPDF(Filename, out PDFPageWidth, out PDFPageHeight);
//break up text objects with multiple words into a single word per object
DataBase.DB DocObjects = new DataBase.DB();
DocObjects.LocateKeywords();
ImageProcessor EvalImage = new ImageProcessor();
int ImageWidth = 0;
int ImageHeight = 0;
List <Rectangle> BlackBoxes = EvalImage.BlackBox(Filename + ".png", out ImageWidth, out ImageHeight, true);
double ImageScale = PDFPageWidth / ImageWidth;
DocObjects.ExtractTextFromRect(BlackBoxes, ImageScale);
List <Rectangle> BlackBlobs = EvalImage.BlackBlob(Filename + ".png", out ImageWidth, out ImageHeight, true);
ImageScale = PDFPageWidth / ImageWidth;
DocObjects.ExtractTextFromRect(BlackBoxes, ImageScale);
DocObjects.TranslateText();
ArchiveTrainingData();
PopulateGridView("Extract");
}
