Marmelator/ocr-service
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 64 Wiki Activity

RED-7669: optimize OCR-module performance

Browse Source 
* second attempt at thread safety
This commit is contained in:
Kilian Schuettler 2023-11-15 15:29:18 +01:00
parent 57e194fcd0
commit 77355b5367
3 changed files with 16 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/model/ExtractedOcrImage.java
View File

@ -45,12 +45,22 @@ public class ExtractedOcrImage implements OcrImage {
this.originalHeight = bufferedImage.getHeight(); this.originalHeight = bufferedImage.getHeight();
this.originalWidth = bufferedImage.getWidth(); this.originalWidth = bufferedImage.getWidth();
float imageDPI = Math.abs(bufferedImage.getWidth() / (ctm.getScalingFactorX() / 72)); float imageDPI = Math.abs(bufferedImage.getWidth() / (ctm.getScalingFactorX() / 72));
this.pix = ImageProcessingUtils.process(bufferedImage, imageDPI, targetDpi); this.pix = binarize(bufferedImage, imageDPI, targetDpi);
this.height = pix.h; this.height = pix.h;
this.width = pix.w; this.width = pix.w;
} }
@SneakyThrows
synchronized private Pix binarize(BufferedImage image, float imageDpi, int targetDpi) {
ImageProcessingUtils.setAlphaChannelToWhite(image);
Pix grayScale = ImageProcessingUtils.convertToGrayScale(image);
Pix scaledUp = ImageProcessingUtils.scaleToTargetDpi(imageDpi, targetDpi, grayScale);
return ImageProcessingUtils.despecklePix(scaledUp);
}
@Override @Override
public AffineTransform getImageCTM() { public AffineTransform getImageCTM() {

25
ocr-service-v1/ocr-service-processor/src/main/java/com/knecon/fforesight/service/ocr/processor/utils/ImageProcessingUtils.java
View File

@ -15,24 +15,7 @@ import net.sourceforge.lept4j.util.LeptUtils;
@UtilityClass @UtilityClass
public class ImageProcessingUtils { public class ImageProcessingUtils {
public Pix process(BufferedImage image, float imageDpi, int targetDpi) { public static Pix despecklePix(Pix scaledUp) {
setAlphaChannelToWhite(image);
return processWithLeptonica(image, imageDpi, targetDpi);
}
// LeptUtils and Leptonica1 does not seem to be thread safe, so we must ensure synchronization for the image processing.
// There might be a way to get this working multi-threaded, but it does not seem to be a significant runtime factor, so i didn't bother investing the time to dive deeper.
synchronized private static Pix processWithLeptonica(BufferedImage image, float imageDpi, int targetDpi) {
Pix grayScale = convertToGrayScale(image);
Pix scaledUp = scaleToTargetDpi(imageDpi, targetDpi, grayScale);
return despecklePix(scaledUp);
}
private static Pix despecklePix(Pix scaledUp) {
assert scaledUp.d == 8; assert scaledUp.d == 8;
Pix despeckled = LeptUtils.despeckle(scaledUp, LeptUtils.SEL_STR3, 3); Pix despeckled = LeptUtils.despeckle(scaledUp, LeptUtils.SEL_STR3, 3);
@ -46,7 +29,7 @@ public class ImageProcessingUtils {
} }
private static Pix scaleToTargetDpi(float imageDpi, int targetDpi, Pix grayScale) { public static Pix scaleToTargetDpi(float imageDpi, int targetDpi, Pix grayScale) {
float targetFactor = targetDpi / imageDpi; float targetFactor = targetDpi / imageDpi;
@ -67,7 +50,7 @@ public class ImageProcessingUtils {
@SneakyThrows @SneakyThrows
private static Pix convertToGrayScale(BufferedImage image) { public static Pix convertToGrayScale(BufferedImage image) {
Pix pix = LeptUtils.convertImageToPix(image); Pix pix = LeptUtils.convertImageToPix(image);
if (pix.d == 8) { if (pix.d == 8) {
@ -84,7 +67,7 @@ public class ImageProcessingUtils {
} }
private static void setAlphaChannelToWhite(BufferedImage image) { public static void setAlphaChannelToWhite(BufferedImage image) {
if (image.getTransparency() == Transparency.TRANSLUCENT) { if (image.getTransparency() == Transparency.TRANSLUCENT) {
// NOTE: For BITMASK images, the color model is likely IndexColorModel, // NOTE: For BITMASK images, the color model is likely IndexColorModel,

1
ocr-service-v1/ocr-service-server/src/test/java/com/knecon/fforesight/service/ocr/v1/server/OcrServiceIntegrationTest.java
View File

@ -172,6 +172,7 @@ public class OcrServiceIntegrationTest extends AbstractTest {
} }
@SneakyThrows @SneakyThrows
private void testOCRForFile(File file) { private void testOCRForFile(File file) {