Added code cleaning up the images extracted for OCR.

Author: klanderfri

ReadMagicCard/OcrImageNoiseCleaner.cpp

@@ -0,0 +1,79 @@
+#include "stdafx.h"
+#include "OcrImageNoiseCleaner.h"
+
+using namespace cv;
+using namespace std;
+
+OcrImageNoiseCleaner::OcrImageNoiseCleaner(wstring imageFileName, Mat originalImageData, SystemMethods* systemMethods, bool runDebugging)
+	: BasicReaderData(imageFileName, originalImageData, systemMethods, runDebugging)
+{
+}
+
+OcrImageNoiseCleaner::~OcrImageNoiseCleaner()
+{
+}
+
+void OcrImageNoiseCleaner::CleanImage(Mat& dirtyImage) {
+
+	Mat originalImage;
+	dirtyImage.copyTo(originalImage);
+
+	threshold(dirtyImage, dirtyImage, 130, 255, THRESH_BINARY);
+	Contours contours = ImageHelper::GetCannyContours(dirtyImage, 120);
+	LetterAreas figures = ImageHelper::ToLetterAreas(contours);
+
+	for (size_t i = 0; i < figures.size(); i++) {
+
+		handleFigure(figures[i], dirtyImage);
+	}
+}
+
+void OcrImageNoiseCleaner::handleFigure(LetterArea figure, Mat& dirtyImage) {
+
+	if (figure.Box.size.area() == 0) { return; }
+
+	Mat contourImage = ImageHelper::DrawLimits(dirtyImage, { figure.TightContour }, Hierarchy(), false);
+
+	if (isNoise(figure, dirtyImage.size())) {
+
+		Scalar white(255, 255, 255);
+		ImageHelper::FillContour(dirtyImage, figure.OuterContour, white);
+	}
+}
+
+bool OcrImageNoiseCleaner::isNoise(LetterArea figure, Size imageArea) {
+
+	if (isOblong(figure)) { return true; }
+	if (isSmallEnoughToBeNoise(figure) &&
+		!isInCenter(figure, imageArea)) { return true; }
+
+	return false;
+}
+
+bool OcrImageNoiseCleaner::isOblong(LetterArea figure) {
+
+	double height = min(figure.Box.size.height, figure.Box.size.width);
+	double width = max(figure.Box.size.height, figure.Box.size.width);
+	double relation = width / height;
+
+	return (relation > 5.5);
+}
+
+bool OcrImageNoiseCleaner::isSmallEnoughToBeNoise(LetterArea figure) {
+
+	bool isOpen = !isContourConvex(figure.TightContour);
+	double area = contourArea(isOpen ? figure.OuterContour : figure.TightContour);
+	int maxNoiseArea = (int)(WORKING_CARD_HEIGHT / 8.5); //80
+
+	return (area <= maxNoiseArea);
+}
+
+bool OcrImageNoiseCleaner::isInCenter(LetterArea figure, Size imageArea) {
+
+	int borderMargin = (int)(WORKING_CARD_HEIGHT / 27.2); //25
+	Rect rCenter(borderMargin, borderMargin, imageArea.width - 2 * borderMargin, imageArea.height - 2 * borderMargin);
+	RotatedRect rrCenter = ImageHelper::ToRotatedRectangle(rCenter);
+	bool placedInCenter = ImageHelper::DoesRectangleContainPoint(rrCenter, figure.Box.center);
+
+	return placedInCenter;
+}

ReadMagicCard/OcrImageNoiseCleaner.h

@@ -0,0 +1,26 @@
+#pragma once
+#include "BasicReaderData.h"
+class OcrImageNoiseCleaner :
+	public BasicReaderData
+{
+public:
+	OcrImageNoiseCleaner(std::wstring imageFileName, cv::Mat originalImageData, SystemMethods* systemMethods, bool runDebugging);
+	~OcrImageNoiseCleaner();
+
+	////Cleans the text images from noise and clutter.
+	void CleanImage(cv::Mat& dirtyImage);
+
+private:
+
+	//Handles a single figure (noise or letter).
+	void handleFigure(LetterArea figure, cv::Mat& dirtyImage);
+	//Checks if a figure is noise.
+	bool isNoise(LetterArea figure, cv::Size imageArea);
+	//Checks if a figure is too long and thin to be a letter.
+	bool isOblong(LetterArea figure);
+	//Checks if a figure is so small it can be considered noise.
+	bool isSmallEnoughToBeNoise(LetterArea figure);
+	//Checks if a figure is in the middle of the text image.
+	bool isInCenter(LetterArea figure, cv::Size imageArea);
+};
+

ReadMagicCard/ReadMagicCard.vcxproj

@@ -189,6 +189,7 @@ xcopy /Y "$(ProjectDir)deployment\vcredist_x86.exe" "$(OutDir)"</Command>
     <ClInclude Include="LoadOcvImage.h" />
     <ClInclude Include="MtgCardInfoHelper.h" />
     <ClInclude Include="OcrDecodeResult.h" />
+    <ClInclude Include="OcrImageNoiseCleaner.h" />
     <ClInclude Include="ReadingConfiguration.h" />
     <ClInclude Include="SaveOcvImage.h" />
     <ClInclude Include="SectionExtractor.h" />
@@ -224,6 +225,7 @@ xcopy /Y "$(ProjectDir)deployment\vcredist_x86.exe" "$(OutDir)"</Command>
     <ClCompile Include="LoadOcvImage.cpp" />
     <ClCompile Include="MtgCardInfoHelper.cpp" />
     <ClCompile Include="OcrDecodeResult.cpp" />
+    <ClCompile Include="OcrImageNoiseCleaner.cpp" />
     <ClCompile Include="ReadingConfiguration.cpp" />
     <ClCompile Include="SaveOcvImage.cpp" />
     <ClCompile Include="SectionExtractor.cpp" />

ReadMagicCard/ReadMagicCard.vcxproj.filters

@@ -168,6 +168,9 @@
     <ClInclude Include="StoreCardProcessingData.h">
       <Filter>Application</Filter>
     </ClInclude>
+    <ClInclude Include="OcrImageNoiseCleaner.h">
+      <Filter>Engine\Helpers</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <ClCompile Include="CardNameInfo.cpp">
@@ -266,6 +269,9 @@
     <ClCompile Include="StoreCardProcessingData.cpp">
       <Filter>Application</Filter>
     </ClCompile>
+    <ClCompile Include="OcrImageNoiseCleaner.cpp">
+      <Filter>Engine\Helpers</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <None Include="tessdata\eng.traineddata">

ReadMagicCard/TitleExtractor.cpp

@@ -5,6 +5,7 @@
 #include "AlgorithmHelper.h"
 #include "SaveOcvImage.h"
 #include "LetterFilter.h"
+#include "OcrImageNoiseCleaner.h"
 
 using namespace cv;
 using namespace std;
@@ -38,6 +39,9 @@ bool TitleExtractor::ExtractTitle(vector<Mat>& outImages, int binaryThreshold, i
 	//Extract a clean image containing the title.
 	bool success = getTitleText(outImage, outImages, numberOfTries);
 
+	//Clean the text images from noise and clutter.
+	cleanOcrImages(outImages);
+
 	//Store that the extration has been run, indicating that we can
 	//extract extra data (such as if the title has black background colour).
 	hasRunTitleExtraction = true;
@@ -133,7 +137,6 @@ bool TitleExtractor::getTitleText(const Mat titleImage, vector<Mat>& textImages,
 	//Cut out the title text.
 	int borderThickness = 10;
 	ImageHelper::CropImageWithSolidBorder(straightenTitleImage, straightenTitleImage, straightTextArea, borderThickness);
-
 	textImages.push_back(straightenTitleImage);
 
 	if (letters.size() < 7) {
@@ -212,3 +215,13 @@ RotatedRect TitleExtractor::getTextArea(Contour letters, TrendLine centerLine, T
 
 	return textArea;
 }
+
+void TitleExtractor::cleanOcrImages(vector<Mat>& outImages) {
+
+	OcrImageNoiseCleaner cleaner(imageFileName, originalImageData, systemMethods, runDebugging);
+
+	for (Mat dirtyImage : outImages) {
+
+		cleaner.CleanImage(dirtyImage);
+	}
+}

ReadMagicCard/TitleExtractor.h

@@ -26,6 +26,8 @@ class TitleExtractor :
 	bool getTitleText(const cv::Mat titleImage, std::vector<cv::Mat>& textImages, int& numberOfTries);
 	//Gets the rectangle bounding the text area.
 	cv::RotatedRect getTextArea(Contour letters, TrendLine centerLine, TrendLine baseLine, cv::Mat titleImage, int numberOfTries);
+	//Cleans the text images from noise and clutter.
+	void cleanOcrImages(std::vector<cv::Mat>& outImages);
 
 	//Tells if the original title image has a black background.
 	bool hasOriginalTitleBlackBackground;