public TextExtractionResults Extract(byte[] image)
{
var end = Amazon.RegionEndpoint.EUWest1;
Amazon.Runtime.BasicAWSCredentials awsCreds =
new Amazon.Runtime.BasicAWSCredentials(this.AmazonAccessKey, this.AmazonSecretKey);
var region = Amazon.RegionEndpoint.GetBySystemName("eu-west-1");
var cfg = new Amazon.Textract.AmazonTextractConfig();
Amazon.Textract.AmazonTextractClient client = new Amazon.Textract.AmazonTextractClient(awsCreds, end);
//Amazon.Textract.AmazonTextractClient client1 = new Amazon.Textract.AmazonTextractClient(awsCreds, cfg);
var request = new Amazon.Textract.Model.DetectDocumentTextRequest();
request.Document = new Amazon.Textract.Model.Document();
Amazon.Textract.Model.DetectDocumentTextResponse result = null;
using (var memstm = new System.IO.MemoryStream(image))
{
request.Document.Bytes = memstm;
result = client.DetectDocumentTextAsync(request).Result;
}
TextExtractionResults rs = new TextExtractionResults();
List <TextBlock> lstTextBoxes = new List <TextBlock>();
DetermineDimensions(image);
foreach (var block in result.Blocks)
{
TextBlock rsText = new TextBlock();
rsText.Text = block.Text;
if (string.IsNullOrWhiteSpace(rsText.Text))
{
continue; //The first element is always NULL, so it appears
}
if (block.BlockType == Amazon.Textract.BlockType.WORD)
{
double xTopLeft = block.Geometry.BoundingBox.Left * this.Width;
double yTopLeft = block.Geometry.BoundingBox.Top * this.Height;
double width = block.Geometry.BoundingBox.Width * this.Width;
double height = block.Geometry.BoundingBox.Height * this.Height;
rsText.X1 = xTopLeft;
rsText.Y1 = yTopLeft;
rsText.X2 = rsText.X1 + width;
rsText.Y2 = rsText.Y1 + height;
lstTextBoxes.Add(rsText);
}
else
{
//Para or line perhaps?
}
}
rs.Blocks = lstTextBoxes.ToArray();
return(rs);
}
private void RenderJson(TextExtractionResults lastOcrResults)
{
try
{
var settings = new Newtonsoft.Json.JsonSerializerSettings
{
Formatting = Newtonsoft.Json.Formatting.Indented,
};
string text = Newtonsoft.Json.JsonConvert.SerializeObject(lastOcrResults, settings);
ctlJsonViewer.Text = "";
ctlJsonViewer.Text = text;
}
catch (Exception ex)
{
MessageBox.Show(ex.ToString());
}
}
public TextExtractionResults Extract(byte[] image)
{
HttpClient httpClient = new HttpClient();
httpClient.Timeout = new TimeSpan(1, 1, 1);
string lang = "eng";
MultipartFormDataContent form = new MultipartFormDataContent();
//form.Add(new StringContent("helloworld"), "apikey"); //This works ! I am not sure how?
form.Add(new StringContent(this.ApiKey), "apikey"); //Added api key in form data
form.Add(new StringContent(lang), "language");
form.Add(new StringContent("True"), "isOverlayRequired");
form.Add(new ByteArrayContent(image, 0, image.Length), "image", "image.jpg");
HttpResponseMessage response = httpClient.PostAsync("https://api.ocr.space/Parse/Image", form).Result;
string strContent = response.Content.ReadAsStringAsync().Result;
entity.Rootobject ocrResult = JsonConvert.DeserializeObject <entity.Rootobject>(strContent);
TextExtractionResults rs = new TextExtractionResults();
entity.Word[] allWords = ocrResult.
ParsedResults.
SelectMany(r => r.TextOverlay.Lines).
SelectMany(ln => ln.Words).
ToArray();
List <TextBlock> lstTextWords = new List <TextBlock>();
foreach (var word in allWords)
{
TextBlock rsTxt = new TextBlock
{
Text = word.WordText,
X1 = word.Left,
Y1 = word.Top
};
rsTxt.X2 = word.Left + word.Width;
rsTxt.Y2 = word.Top + word.Height;
lstTextWords.Add(rsTxt);
}
rs.Blocks = lstTextWords.ToArray();
return(rs);
}
public TextExtractionResults Extract(byte[] image)
{
TextExtractionResults rs = new TextExtractionResults();
List <TextBlock> rsTextBoxes = new List <TextBlock>();
AmazonRekognitionClient rekognitionClient = CreateAwsClient();
using (var stm = new System.IO.MemoryStream(image))
{
double imgWidth, imgHeight;
using (var bmp = new System.Drawing.Bitmap(stm))
{
imgWidth = bmp.Width;
imgHeight = bmp.Height;
}
var awsImage = new Amazon.Rekognition.Model.Image();
awsImage.Bytes = new System.IO.MemoryStream(image);
var req = new Amazon.Rekognition.Model.DetectTextRequest
{
Image = awsImage
};
var detectTextResponse = rekognitionClient.DetectTextAsync(req).Result;
foreach (Amazon.Rekognition.Model.TextDetection textResult in detectTextResponse.TextDetections)
{
string text = textResult.DetectedText;
Amazon.Rekognition.Model.Geometry oGeom = textResult.Geometry;
TextBlock rsText = new TextBlock();
rsText.Text = text;
rsText.X1 = oGeom.BoundingBox.Left * imgWidth;
rsText.X2 = rsText.X1 + oGeom.BoundingBox.Width * imgWidth;
rsText.Y1 = oGeom.BoundingBox.Top * imgHeight;
rsText.Y2 = rsText.Y1 + oGeom.BoundingBox.Height * imgHeight;
rsTextBoxes.Add(rsText);
}
}
rs.Blocks = rsTextBoxes.ToArray();
return(rs);
}
private void RenderImage(byte[] raw, TextExtractionResults lastOcrResults)
{
Contracts.entity.TextExtractionResults ocr = null;
if (lastOcrResults == null)
{
ctlStatusPanel0.Text = $"Found {lastOcrResults.Blocks.Length} text objects";
//Create an empty object if none was specified
ocr = new Contracts.entity.TextExtractionResults
{
Blocks = new Contracts.entity.TextBlock[]
{
}
};
}
else
{
ocr = lastOcrResults;
}
System.Drawing.Pen penBlock = new System.Drawing.Pen(System.Drawing.Color.Black);
System.Drawing.Pen penSentence = new System.Drawing.Pen(System.Drawing.Color.Orange, 3);
System.Drawing.Pen penPara = new System.Drawing.Pen(System.Drawing.Color.Blue, 3);
System.Drawing.Brush b = new System.Drawing.SolidBrush(
System.Drawing.Color.FromArgb(100, System.Drawing.Color.Yellow));
using (var memStm = new System.IO.MemoryStream(raw))
{
var imge = System.Drawing.Image.FromStream(memStm);
_picBox.Image = imge;
using (var g = System.Drawing.Graphics.FromImage(imge))
{
foreach (var box in ocr.Blocks)
{
//var pts = new System.Drawing.Point[]
//{
// new System.Drawing.Point((int)box.X1,(int)box.Y1),
// new System.Drawing.Point((int)box.X2,(int)box.Y1),
// new System.Drawing.Point((int)box.X2,(int)box.Y2),
// new System.Drawing.Point((int)box.X1,(int)box.Y2),
// new System.Drawing.Point((int)box.X1,(int)box.Y1)
//};
//g.DrawLines(pen,pts);
int xUpperLeft = (int)Math.Min(box.X1, box.X2);
int yUpperLeft = (int)Math.Min(box.Y1, box.Y2);
float width = (float)Math.Abs(box.X1 - box.X2);
float ht = (float)Math.Abs(box.Y1 - box.Y2);
g.DrawRectangle(penBlock, xUpperLeft, yUpperLeft, width, ht);
g.FillRectangle(b, xUpperLeft, yUpperLeft, width, ht);
}
foreach (var sentence in ocr.Sentences)
{
g.DrawLine(
penSentence,
sentence.Rectangle.X, sentence.Rectangle.Bottom,
sentence.Rectangle.Right, sentence.Rectangle.Bottom);
}
foreach (var para in ocr.Paragraphs)
{
g.DrawRectangle(
penPara,
para.Rectangle.Left, para.Rectangle.Top,
para.Rectangle.Width, para.Rectangle.Height);
}
}
}
}
public TextExtractionResults ExtractEntitiesFromJson(string jsonText)
{
//TextExtractionResults rs = new TextExtractionResults();
//JObject json = JObject.Parse(jsonText);
//IEnumerable<JToken> words = json.SelectTokens("$..words");
//JToken[] arr = words.ToArray();
//JToken[] arrAllChildren = arr.SelectMany(t => t).ToArray();
//rs.Blocks = arrAllChildren.
// Select(r => CreateFromJtoken(r)).
// ToArray();
TextExtractionResults rs = new TextExtractionResults();
JObject json = JObject.Parse(jsonText);
List <Paragraph> lstParas = new List <Paragraph>();
IEnumerable <JToken> sections = json.
SelectTokens("$..regions").
SelectMany(l => l).ToArray();
foreach (JToken tSection in sections)
{
string boxCoordinates = tSection["boundingBox"].Value <string>();
float[] dblCoordinates =
boxCoordinates.Split(',').
Select(frag => float.Parse(frag)).
ToArray();
Paragraph paraNew = new Paragraph();
var rect = new System.Drawing.RectangleF(
dblCoordinates[0], dblCoordinates[1],
dblCoordinates[2], dblCoordinates[3]);
paraNew.Rectangle = rect;
lstParas.Add(paraNew);
}
rs.Paragraphs = lstParas.ToArray();
///
/// Use the sentences approach - May ,2019
///
IEnumerable <JToken> lines = json.
SelectTokens("$..lines").
SelectMany(l => l).ToArray();
List <TextBlock> lstAllBlocks = new List <TextBlock>();
List <Sentence> lstAllSentences = new List <Sentence>();
foreach (JToken line in lines)
{
string boxCoordinates = line["boundingBox"].Value <string>();
float[] dblCoordinates =
boxCoordinates.Split(',').
Select(frag => float.Parse(frag)).
ToArray();
Sentence sentNew = new Sentence();
sentNew.Rectangle = new System.Drawing.RectangleF(
dblCoordinates[0], dblCoordinates[1],
dblCoordinates[2], dblCoordinates[3]);
JToken[] arrAllWords = line.
SelectTokens("$..words").
SelectMany(t => t).ToArray();
TextBlock[] blockFromWords = arrAllWords.
Select(r => CreateFromJtoken(r)).
ToArray();
sentNew.Blocks = blockFromWords;
lstAllSentences.Add(sentNew);
lstAllBlocks.AddRange(blockFromWords);
}
rs.Sentences = lstAllSentences.ToArray();
rs.Blocks = lstAllBlocks.ToArray();
return(rs);
}
