Version 10.0.0

◼Added: Detection of Aruco markers and Charuco chessboard corners
◼Added: Contrast Limited Adaptive Histogram Equalization (CLAHE) for local contrast enhancement of gray and color images
◼Added: Poisson seamless cloning to blend a masked source into an image
◼Added: Shi-Tomasi and Harris corner detection
◼Added: JPEG Turbo Lib compiled into IEVision, for 2-6x faster loading using SIMD
◼Added: Generation and detection of QR codes
◼Added: OCR preprocessing in one call: normalizes the background, enlarges small text, corrects skew, binarizes and cleans up document images for more accurate OCR
◼Added: Sauvola local adaptive binarization for documents with uneven lighting
◼Added: Background normalization to flatten uneven lighting and shadows
◼Added: Document deskew and skew measurement
◼Added: Connected-component analysis, with speck and border removal
◼Added: mapRectToSource maps OCR boxes found in a prepared image back to the original image
◼Added: Neural network face detection (YuNet) finds tilted, turned and small faces more reliably than Haar cascades, and returns five facial landmarks (eyes, nose tip, mouth corners) for each face
◼Added: alignFace returns upright, uniformly framed face crops for more reliable face recognition
◼Added: Accurate landmarks for tilted faces, and detection of sideways and upside down faces
◼Added: Face helpers to level a photo by the eyes (optionally trimming the rotated corners), crop to the main face (padding, headroom and aspect ratio) and pixelize faces to anonymize photos
◼Added: convexHull and convexHullIndices for shape analysis
◼Added: Delaunay triangulation for mesh warping and morphing
◼Added: Affine transformations with getAffineTransform and warpAffine
◼Added: Polygon filling with fillConvexPoly and fillPoly, and createVectorObjRef to create polygon lists
◼Added: Perceptual image hashing (pHash, average, block mean, Marr-Hildreth, radial variance and color moment) to find similar and near-duplicate images, with GetPerceptualHash, IEVisionCompareHashes, FindSimilar and near-duplicate removal with RemoveDuplicates
◼Added: TIEVisionFeaturesDetector supports SIFT, AKAZE, KAZE and BRISK as well as ORB, with Lowe's ratio test and cross-checking for more reliable matching and image alignment (TIEVisionORBFeaturesDetector is now an alias)
◼Added: Removal of form and table lines and underlines before OCR (also available as the ievOCRPrepRemoveLines step of prepareForOCR), and getLineMask

Version 9.0.0

◼Added: ZBar 0.10 upgraded to 0.23.93 adding support for GS1-DataBar, Code-93 and Codabar
◼Added: ZBar: Security and memory-safety fixes, including CVE-2023-40889
◼Added: Boost upgraded from 1.69.0 to 1.92.0 for reliability and security improvements
◼Added: Boost: Reliability/security fixes and improved modern compiler/platform support
◼Added: Upgrade OpenCV from 4.6.0 to 4.14.0 for reliability, performance and security improvements
◼Added: OpenCV: Accumulated fixes for undefined behaviour, integer overflow and memory-safety issues, among many other fixes
◼Fixed: Updated Leptonica 1.74.2 to 1.87.0 and Tesseract 5.3.3 to 5.5.3
◼Added: LibRaw upgraded from 0.20 to 0.22.2 for security fixes for OCR accuracy improvements
◼Added: LibRaw: Improved validation and protection against malformed RAW input, including size/index and integer-overflow checks

Version 8.3.1

◼Fixed: Exception of IEVision library is unloaded and reloaded in secondary DLLs

Version 8.3.0

◼Added: houghCircles method
◼Added: Min and max line angle detection added to houghLines
◼Fixed: Specifying an invalid rect for OCR may cause an exception
◼Added: Calculation of the blurriness of images using Laplacian variance
◼Added: Updated LibRaw to 0.22.0

Version 8.1.6

◼Added: Updated LibRaw to 0.21.4
◼Added: Various enhancements to template matching (image searching) to improve result quality

Version 8.1.5

◼Added: Further improvements to calculation of orientation and angle in documents

Version 8.1.4

◼Added: thinPlateSplineShapeTransform supports replication when filling border areas, as well as custom colors
◼Added: Much improved algorithm for detecting the text angle and orientation in documents

Version 8.1.3

◼Added: New method to check successful language loading for Searchable PDF generator
◼Added: Improvements to filter2D

Version 8.1.1

◼Fixed: Unable to output in JPEG format with TIEVisionSearchablePDFGenerator
◼Fixed: Saving a PDF with TIEVisionSearchablePDFGenerator using the same name may fail

Version 8.1.0

◼Added: Support for a local model neural network to classify objects, detect text and increase resolution
◼Added: Searchable PDF documents can now be created without images
◼Added: New HAAR classifiers: Eye pair, Nose, Mouth, Stop Sign
◼Added: New method to check successful language loading for OCR
◼Added: Improved exception messages for OCR

Version 8.0.0

◼Added: OCR now built on Tesseract 5 for greater performance using less system memory and improved accuracy, particularly with images of low resolution or contrast
◼Added: OCR results can be returned in hOCR, ALTO (XML) and text-separated value formats
◼Added: Support for Black and White lists with OCR
◼Added: New OCR constructor for analysis only
◼Added: Confidence values returned for OCR results
◼Added: Many enhancements to stitching including new thresholding options and improved performance with documents
◼Added: Prevents usage of invalid parameters when smoothing

Version 7.1.0

◼Added: Thin Plate Spline shape transformations
◼Fixed: Some OCR issues

Version 7.0.0

◼Added: Updated OpenCV to v4.6
◼Added: Updated LibRaw to v0.21.1 with support for newer Camera formats
◼Added: Implemented LibJpeg-Turbo
◼Added: New "Scans" option for image stitcher
◼Added: Multiple matches can now be found with matchTemplateMulti
◼Fixed: Support for RGB images minMax()
◼Fixed: Searchable PDF Generator now uses Windows temp path for temporary files
◼Added: Highly optimized DLLs. Files are now nearly 50% smaller

Version 6.0.4

◼Fixed: Possible exception when using Binary Descriptor line detection with video feed
◼Fixed: Error when using very different images with findHomography
◼Fixed: No image change using IEImage.smooth(), pyrUp(), pyrDown() without destination bitmap

Version 6.0.3

◼Added: support CgBI PNG (Apple's proprietary extension to PNG)
◼Fixed: Error when using Canny edge detection
◼Added: Support for JPEG with arithmetic encoding

Version 6.0.1

◼Added: Compression can be configured for searchable PDF generator
◼Fixed: Some exceptions that occur under 64bit
◼IELib: Updated OpenJPEG to 2.4.0
◼IELib: Added support for X3F and 8bit DNG raw images

Version 6.0.0

◼Added: Orb Feature Detector class to find matching points in a pair of images
◼Added: FindHomography method to find the alignment matrix between points in a pair of images (e.g. for automatic image alignment)
◼Added: WarpPerspective method to adjust the perspective of an image based on a matrix
◼Added: Exposure Fusion class: Merge images taken with multiple exposures into a single image
◼Added: Updated all OCR language files and added OCR script files (e.g. all Latin languages)
◼Added: Support for UTF encoding reading barcodes
◼Fixed: PDF generator may produce inverted images

Version 5.5.0

◼Fixed: Loading NEF files give Invalid Floating point error (64bit)
◼Added: Three new OCR page segmentation modes
◼Fixed: Handles leak using OCR
◼OCR: Updated to Tesseract 4.1.1

Version 5.4.2

◼Fixed: With Searchable PDF, monochrome images output is inverted
◼Jpeg2000: Updated to OpenJpeg 2.3.1
◼LibRaw: Updated to LibRaw 0.20.2
◼Fixed: Background subtractor MOG2 doesn't work for 64 bits
◼New Demo: Subtract Background from Image
◼Fixed: Installer fails to use correct destination folders

Version 5.4.1

◼SearchablePDF: support for color images
◼(IELib) support for 12 bit jpegs (DICOM)

Version 5.4.0

◼Added: Updated LibRaw to 0.20 for EXIF support and newer cameras
◼PDF: support for DPI specification

Version 5.3.0

◼Added: OpenMP (tesseract only) optional DLLs

Version 5.2.0

◼Added: Image Stitcher can now run deterministically
◼Fixed: CreateOCR may crash if path is not explicitly specified
◼Added: Threshold method now returns the calculated thresholding value
◼Added: Uses RandomizedBaseAddress (ASLR) and DataExecutionPrevention (DEP) flags for security
◼New Demo: Recognize faces in a video stream
◼New Demo: Blur any car number plates found in an image

Version 5.1.1

◼Added: Updated LibRaw to 201910 for updated camera support

Version 5.1.0

◼Added: Optimization of JPEG2000 Engine

Version 5.0.0

◼Added: Supports Tesseract 4.0.0 for much faster and more accurate OCR (Error Reduction: English: -30%, French: -50%, Spanish: -46%, Russian: -79%, Hebrew: -32%, Chinese: -22%, Korean -79%, etc.)
◼Added: Support for 60 Languages added to OCR, such as: Armenian, Breton, Corsican, Filipino, Gaelic, Mongolian, Maori, Tongan
◼Added: TIEVisionSearchablePDFGenerator class to generate searchable (OCRed) PDF documents from images
◼Added: TIEVisionOCRWordBox.isWordFromDictionary method
◼Added: TIEVisionOCRWordBox.isSuperscript method
◼Added: TIEVisionOCRWordBox.isSubscript method
◼Added: TIEVisionOCRWordBox.isDropCap
◼Added: Supports LibRaw 0.19.2

Version 4.5.2

◼Added: TIEVisionBlobDetector, support for BW images
◼Fixed: TIEVisionBlobDetector, crash on unsupported formats

Version 4.5.1

◼Added: TIEVisionBarCodeSymbol.getQuality method
◼Fixed: createCascadeClassifier fails under 64bit
◼Fixed: getWordBoxes may raise exception on invalid selection

Version 4.5.0

◼Added: TIEVisionGrabCut.setSelectionMask optional maskIsInside parameter
◼Updated: LibRaw to version 0.18.11
◼Added: TIEVisionBackgroundSubtractor.selectMethodMOG method
◼Added: TIEVisionBackgroundSubtractor.selectMethodGMG method
◼Added: TIEVisionImage.detectLines method
◼Added: TIEVisionImage.matchTemplate optional rank parameter
◼Added TIEVisionBlobDetector class
◼Added TIEVisionLibrary.createBlobDetector method

Version 4.0.0

◼Added: TIEVisionImage.matchTemplate to perform template matching
◼Added: TIEVisionImage.matchTemplateAsMap to perform template matching with results map
◼Added: TIEVisionCascadeClassifierTrainer class to train (HAAR or LBP) cascade classifiers for object detection
◼Added: TIEVisionGrabCut class to automatically select regions
◼Added: TIEVisionWatershed class to automatically select regions

Version 3.0.8

◼Fixed: inPaint crash when brush size is smaller than ROI rectangle

Version 3.0.7

◼Fixed: Lossless transformations do not work
◼Added: Improved jpeg2000 loading and saving speed

Version 3.0.6

◼Updated dcraw to version 1.477 (9.27)

Version 3.0.5

◼Changed: findIn limits maximum size to 2 MegaPixels, to avoid out of memory errors (can still happen when scaleFactor is less than 1.1)
◼Removed: OpenCL support to avoid problems on some CPUs

Version 3.0.4

◼Added: FaceRecognizer adds clear, isTrained, train, load and save overloads
◼Added: ObjectsFinder.setHaarMaxSize and setHaarMinSize methods
◼Added: ObjectsFinder.setEqualizeHistogram method to make histogram equalization optional

Version 3.0.3

◼Updated dcraw to version 1.476 (9.26)
◼Added: Linked dcraw to jpeglib and jasperlib to support some Kodak devices

Version 3.0.2

◼Fixed: Issue loading some reduced size jpegs

Version 3.0.1

◼Fixed: OCR issue due to disabled inline expansion
◼Fixed: OCR environment path problem

Version 3.0.0

◼Added: New edge detection algorithm in TIEVisionImage.canny
◼Added: Line and segments detection in TIEVisionImage.houghLines and TIEVisionImage.houghLinesP
◼Added: New blur algorithm in TIEVisionImage.blur
◼Added: Panorama stitching in TIEVisionStitcher class
◼Added: Image denoising using Non-local Means Denoising algorithm TIEVisionImage.fastNlMeansDenoisingColored
◼Added: New thresholding algorithms in TIEVisionImage.threshold and TIEVisionImage.adaptiveThreshold
◼Added: New erode/dilate algorithms in TIEVisionImage.erode and TIEVisionImage.dilate
◼Added: Background subtractor in TIEVisionBackgroundSubtractor class
◼Added: Face recognizer in TIEVisionFaceRecognizer class
◼Added: new IEVision detectable objects IEVC_CASCADE_SMILE, IEVC_FRONTAL_CAT_FACE, IEVC_FRONTAL_CAT_FACE_EXT, IEVC_LICENSE_PLATE_RUS, IEVC_RUSSIAN_PLATE_NUMBER
◼Fixed: Wrong bitmap orientation using IEVision and TImageEnProc.Undo
◼Removed: IEVision detectable objects: IEVC_MCS_EYE_PAIR_BIG, IEVC_MCS_EYE_PAIR_SMALL, IEVC_MCS_LEFT_EYE, IEVC_MCS_MOUTH, IEVC_MCS_NOSE, IEVC_MCS_RIGHT_EYE, IEVC_MCS_UPPER_BODY, IEVC_MCS_LEFT_EAR, IEVC_MCS_RIGHT_EAR

Version 2.0.2

◼Added: OCR MICR reading support

Version 2.0.1

◼Updated DCRAW to support newest Digital Camera Raw formats
◼Added: Barcode decoder now supports monochrome images

Version 2.0.0

◼Added: Barcode reading support (EAN-13/UPC-A, UPC-E, EAN-8, Code 128, Code 39, Interleaved 2 of 5 and QR Code)
◼Added: TIEVisionBarCodeSymbol class
◼Added: TIEVisionBarCodeScanner class
◼Added: TIEVisionLibrary.createBarCodeScanner method
◼Added: TIEVisionLibrary.getLibraryInfo method

Version 1.0.4

◼Updated OCR library (now supports more languages with enhanced accuracy)
◼Changed: With OCR, English is no longer embedded
◼Changed: createOCR accepts only language code instead of file name
◼Added: OCR now supports multithreading
◼Removed: TIEVisionOCR.setAccuracy method
◼Added: TIEVisionOCR.getRegions, added onlyText parameter to get only text regions
◼Added: New methods for TIEVisionOCRWordBox: getConfidence, isBold, isItalic, isUnderlined, isMonospace, isSerif, isSmallCaps, getPointSize, getLanguage and getScriptDirection
◼Added: OCR now supports loading and detecting multiple languages (with the same TIEVisionOCR object)

---