From 0cf867b97c2a240f2bbdca479ebdc36f64ed3bbb Mon Sep 17 00:00:00 2001 From: deiflaender Date: Fri, 17 Feb 2023 12:13:39 +0100 Subject: [PATCH] RED-6164: Fixed calculation of image is ocr on scanned pages with cv analysis found tables --- .../v1/server/segmentation/PdfSegmentationService.java | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java index 10388bd4..be16069b 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/PdfSegmentationService.java @@ -101,15 +101,16 @@ public class PdfSegmentationService { page.setPageWidth(cropbox.getWidth()); page.setPageHeight(cropbox.getHeight()); - tableExtractionService.extractTables(cleanRulings, page); - buildPageStatistics(page); - increaseDocumentStatistics(page, document); - + // If images is ocr needs to be calculated before textBlocks are moved into tables, otherwise findOcr algorithm needs to be adopted. if (pdfImages != null && pdfImages.containsKey(pageNumber)) { page.setImages(pdfImages.get(pageNumber)); imageService.findOcr(page); } + tableExtractionService.extractTables(cleanRulings, page); + buildPageStatistics(page); + increaseDocumentStatistics(page, document); + pages.add(page); }