Optical Character Recognition (OCR) turns images of text—scans, smartphone photos, PDFs—into machine-readable strings and, increasingly, structured data. Modern OCR is a pipeline that cleans an image, finds text, reads it, and exports rich metadata so downstream systems can search, index, or extract fields. Two widely used output standards are hOCR, an HTML microformat for text and layout, and ALTO XML, a library/archives-oriented schema; both preserve positions, reading order, and other layout cues and are supported by popular engines like Tesseract.
Preprocessing. OCR quality starts with image cleanup: grayscale conversion, denoising, thresholding (binarization), and deskewing. Canonical OpenCV tutorials cover global, adaptive and Otsu thresholding—staples for documents with nonuniform lighting or bimodal histograms. When illumination varies within a page (think phone snaps), adaptive methods often outperform a single global threshold; Otsu automatically picks a threshold by analyzing the histogram. Tilt correction is equally important: Hough-based deskewing (Hough Line Transform) paired with Otsu binarization is a common and effective recipe in production preprocessing pipelines.
Detection vs. recognition. OCR is typically split into text detection (where is the text?) and text recognition (what does it say?). In natural scenes and many scans, fully convolutional detectors like EAST efficiently predict word- or line-level quadrilaterals without heavy proposal stages and are implemented in common toolkits (e.g., OpenCV’s text detection tutorial). On complex pages (newspapers, forms, books), segmentation of lines/regions and reading order inference matter:Kraken implements traditional zone/line segmentation and neural baseline segmentation, with explicit support for different scripts and directions (LTR/RTL/vertical).
Recognition models. The classic open-source workhorse Tesseract (open-sourced by Google, with roots at HP) evolved from a character classifier into an LSTM-based sequence recognizer and can emit searchable PDFs, hOCR/ALTO-friendly outputs, and more from the CLI. Modern recognizers rely on sequence modeling without pre-segmented characters. Connectionist Temporal Classification (CTC) remains foundational, learning alignments between input feature sequences and output label strings; it’s widely used in handwriting and scene-text pipelines.
In the last few years, Transformers reshaped OCR. TrOCR uses a vision Transformer encoder plus a text Transformer decoder, trained on large synthetic corpora then fine-tuned on real data, with strong performance across printed, handwritten and scene-text benchmarks (see also Hugging Face docs). In parallel, some systems sidestep OCR for downstream understanding: Donut (Document Understanding Transformer) is an OCR-free encoder-decoder that directly outputs structured answers (like key-value JSON) from document images (repo, model card), avoiding error accumulation when a separate OCR step feeds an IE system.
If you want batteries-included text reading across many scripts, EasyOCR offers a simple API with 80+ language models, returning boxes, text, and confidences—handy for prototypes and non-Latin scripts. For historical documents, Kraken shines with baseline segmentation and script-aware reading order; for flexible line-level training, Calamari builds on the Ocropy lineage (Ocropy) with (multi-)LSTM+CTC recognizers and a CLI for fine-tuning custom models.
Generalization hinges on data. For handwriting, the IAM Handwriting Database provides writer-diverse English sentences for training and evaluation; it’s a long-standing reference set for line and word recognition. For scene text, COCO-Text layered extensive annotations over MS-COCO, with labels for printed/handwritten, legible/illegible, script, and full transcriptions (see also the original project page). The field also relies heavily on synthetic pretraining: SynthText in the Wild renders text into photographs with realistic geometry and lighting, providing huge volumes of data to pretrain detectors and recognizers (reference code & data).
Competitions under ICDAR’s Robust Reading umbrella keep evaluation grounded. Recent tasks emphasize end-to-end detection/reading and include linking words into phrases, with official code reporting precision/recall/F-score, intersection-over-union (IoU), and character-level edit-distance metrics—mirroring what practitioners should track.
OCR rarely ends at plain text. Archives and digital libraries prefer ALTO XML because it encodes the physical layout (blocks/lines/words with coordinates) alongside content, and it pairs well with METS packaging. The hOCR microformat, by contrast, embeds the same idea into HTML/CSS using classes like ocr_line and ocrx_word, making it easy to display, edit, and transform with web tooling. Tesseract exposes both—e.g., generating hOCR or searchable PDFs directly from the CLI (PDF output guide); Python wrappers like pytesseract add convenience. Converters exist to translate between hOCR and ALTO when repositories have fixed ingestion standards—see this curated list of OCR file-format tools.
The strongest trend is convergence: detection, recognition, language modeling, and even task-specific decoding are merging into unified Transformer stacks. Pretraining on large synthetic corpora remains a force multiplier. OCR-free models will compete aggressively wherever the target is structured outputs rather than verbatim transcripts. Expect hybrid deployments too: a lightweight detector plus a TrOCR-style recognizer for long-form text, and a Donut-style model for forms and receipts.
Tesseract (GitHub) · Tesseract docs · hOCR spec · ALTO background · EAST detector · OpenCV text detection · TrOCR · Donut · COCO-Text · SynthText · Kraken · Calamari OCR · ICDAR RRC · pytesseract · IAM handwriting · OCR file-format tools · EasyOCR
Optical Character Recognition (OCR) is a technology used to convert different types of documents, such as scanned paper documents, PDF files or images captured by a digital camera, into editable and searchable data.
OCR works by scanning an input image or document, segmenting the image into individual characters, and comparing each character with a database of character shapes using pattern recognition or feature recognition.
OCR is used in a variety of sectors and applications, including digitizing printed documents, enabling text-to-speech services, automating data entry processes, and assisting visually impaired users to better interact with text.
While great advancements have been made in OCR technology, it isn't infallible. Accuracy can vary depending upon the quality of the original document and the specifics of the OCR software being used.
Although OCR is primarily designed for printed text, some advanced OCR systems are also able to recognize clear, consistent handwriting. However, typically handwriting recognition is less accurate because of the wide variation in individual writing styles.
Yes, many OCR software systems can recognize multiple languages. However, it's important to ensure that the specific language is supported by the software you're using.
OCR stands for Optical Character Recognition and is used for recognizing printed text, while ICR, or Intelligent Character Recognition, is more advanced and is used for recognizing hand-written text.
OCR works best with clear, easy-to-read fonts and standard text sizes. While it can work with various fonts and sizes, accuracy tends to decrease when dealing with unusual fonts or very small text sizes.
OCR can struggle with low-resolution documents, complex fonts, poorly printed texts, handwriting, and documents with backgrounds that interfere with the text. Also, while it can work with many languages, it may not cover every language perfectly.
Yes, OCR can scan colored text and backgrounds, although it's generally more effective with high-contrast color combinations, such as black text on a white background. The accuracy might decrease when text and background colors lack sufficient contrast.
YUV is a color encoding system used as a part of a color image pipeline. It encodes a color image or video taking human perception into account, allowing reduced bandwidth for chrominance components, thereby typically enabling transmission errors or compression artifacts to be more efficiently masked by the human perception than using a "direct" RGB-representation. The name YUV itself is derived from the Y'UV notation originally used for the luma (Y') and two chrominance (UV) components. The Y'UV model defines a color space in terms of one luma component (Y') and two chrominance components, called U (blue projection) and V (red projection), while YCbCr is a digital version of the Y'UV color model.
YUV signals are created from an original RGB (red, green and blue) source. The weighted values of R, G and B are added together to produce a single Y signal, representing the overall brightness, or luma, of that pixel. The U signal is then created by subtracting the Y from the blue signal of the original RGB, and then scaling; and V by subtracting the Y from the red, and then scaling by a different factor. These factors are chosen to make sure the range of each color space coordinate is roughly -0.5 to +0.5.
The transformation RGB→YUV is specified as follows: Y = 0.299R + 0.587G + 0.114B, U = −0.147R − 0.289G + 0.436B, V = 0.615R − 0.515G − 0.100B. Digital formats commonly use 8 bits for each channel, making the range for each 0 to 255, and so the transform becomes: Y = (0.257 × R) + (0.504 × G) + (0.098 × B) + 16, Cb = U = −(0.148 × R) − (0.291 × G) + (0.439 × B) + 128, Cr = V = (0.439 × R) − (0.368 × G) − (0.071 × B) + 128.
The YUV color model is used in the PAL, NTSC, and SECAM composite color video standards. The luma component is often denoted as Y', but sometimes as Y, prime symbols are often omitted in writing. The YUV system allows the transmission of color images over a channel intended for black-and-white (luma) signals, reducing the bandwidth needed. The black-and-white receivers still display a normal black-and-white picture, while color receivers reverse the process, decoding the UV portions of the signal and displaying a color picture.
One major advantage of YUV is that some of the information may be discarded in order to reduce bandwidth or when chroma is to be processed separately from luma. If only luma needs to be transmitted, that is, the U and V components are zero throughout the frame, then the data size is half of what it was before with no loss to perceived image quality. When converting from full color to YUV and back again, there is some loss of information due to rounding errors.
YUV subsampling is a method of encoding images by implementing less resolution for chroma information than for luma information, taking advantage of the human visual system's lower acuity for color differences than for luminance. 4:4:4 full-resolution YUV stores no chroma subsampling, while common schemes are 4:2:2 (half resolution horizontally), 4:2:0 (half resolution horizontally and vertically) and 4:1:1 (one quarter resolution horizontally). 4:4:4 subsampling preserves all the information present in the original sample. The ratios describe how many luma and chroma samples are encoded for a block of pixels.
There are several shades of YUV color spaces used in video and digital photography systems. The main differences are the scale factors for the U and V planes in the basic equations. While the Y plane represents luminance, and thus requires higher bandwidth, the U and V planes can be bandwidth-reduced, subsampled, compressed, or otherwise treated separately for improved system efficiency. Thus there are several YUV formats, possibly using shades of 8-bit or 10-bit encoding for the planes.
The YUV color model has seen widespread use in digital video, including use in television standards like PAL, NTSC and SECAM, in MPEG compression, in modern digital video interfaces like HDMI, digital video compression schemes like H.264 and VP9, and common image/video container formats such as JPEG/JFIF, PNG and WebP. Its popularity is due to its usefulness in color compression and its ability to take advantage of human perception for more efficient storage and transmission. Overall, YUV remains one of the most important and widely used color models in digital imaging and video.
This converter runs entirely in your browser. When you select a file, it is read into memory and converted to the selected format. You can then download the converted file.
Conversions start instantly, and most files are converted in under a second. Larger files may take longer.
Your files are never uploaded to our servers. They are converted in your browser, and the converted file is then downloaded. We never see your files.
We support converting between all image formats, including JPEG, PNG, GIF, WebP, SVG, BMP, TIFF, and more.
This converter is completely free, and will always be free. Because it runs in your browser, we don't have to pay for servers, so we don't need to charge you.
Yes! You can convert as many files as you want at once. Just select multiple files when you add them.