Optical character recognition API for Python

Extract text from scans, photos and screenshots

Convert scanned PDF to searchable and indexable PDFs

Read Latin, Cyrillic and Asian scripts in all popular typefaces

Recognize more than 6,000 Chinese characters

Extract text from handwritten notes, memos and medical records

Work with blurry, distorted, rotated, and noisy images

Identify and preserve content blocks from the original text layout

Detect and read areas of an image: paragraphs, lines and single words

Bulk process multipage documents, folders and archives

Recognize images from the Internet without downloading

Identify and automatically correct misspelled words

Find text in images and intelligently compare image texts

Easy to use

You only need 5 lines of code to recognize the image and display the result. Yes, it really is that simple!

# Initialize OCR engine
recognitionEngine = AsposeOcr()
# Add image to batch
input = OcrInput(InputType.SINGLE_IMAGE)
input.add("sample.png")
# Extract text from image
result = recognitionEngine.recognize(input)
# Display the recognition result
print(result[0].recognition_text)

28 recognition languages

Aspose.OCR for Python via .NET can recognize a large number of languages and all popular writing scripts, including texts with mixed languages:

• Extended Latin alphabet: Croatian, Czech, Danish, Dutch, English (including handwritten script), Estonian, Finnish, French, German, Italian, Latvian, Lithuanian, Norwegian, Polish, Portuguese, Romanian, Slovak, Slovene, Spanish, Swedish.
• Cyrillic alphabet: Belorussian, Bulgarian, Kazakh, Russian, Serbian, Ukrainian.
• Chinese: more than 6,000 characters.
• Hindi

You can also read texts in other languages based on extended Latin and Cyrillic, even if they are not directly supported by the OCR engine. For example, Latin, Vietnamese, Gaelic, and so on.

Powerful processing filters

The accuracy and reliability of optical character recognition is highly dependent on the quality of the original image. Aspose.OCR for Python via .NET offers a large number of fully automated and manual image processing filters that enhance an image before it is sent to the OCR engine:

• Automatically straighten images aligned at a slight angle to the horizontal.
• Manually rotate severely skewed images.
• Automatically remove dirt, spots, scratches, glare, unwanted gradients, and other noise.
• Automatically adjust the image contrast.
• Automatically upscale, or manually resize the image.
• Convert images to black and white or grayscale.
• Invert image colors so that light areas appear dark and dark areas appear light.
• Increase the thickness of characters in an image.
• Blur noisy images while preserving the edges of letters.
• Straighten page curvature and fix camera lens distortion for page photos.

These filters can be combined and applied to the entire image or only to selected areas of the image, as well as in batch processing. Not only can you fine-tune preprocessing in the recognition pipeline, but you can also store processed images for display, caching, and debugging.

Optimized for specific document types

Aspose.OCR for Python via .NET offers specially trained neural networks to extract text from certain types of images with maximum accuracy:

• Scanned or photographed ID cards and passports.
• Vehicle license plates.
• Invoices.
• Receipts.

Built-in spell checker

Although Aspose.OCR for Python via .NET provides high recognition accuracy, printing defects, dirt, or non-standard fonts may cause certain characters or words to be recognized incorrectly. To further improve recognition results, you can turn on spell checker, which finds and automatically corrects spelling errors based on the selected recognition language.

If the recognized text contains specialized terminology, abbreviations, and other words which are not present in common spelling dictionaries, you can provide your own word lists.

Creating searchable PDFs

Even with the highest recognition accuracy, the original image may contain a lot of important non-textual information or simply be of great historical value. Aspose.OCR for Python via .NET offers a simple and elegant solution to combine the best of both worlds. We extract text from an image, PDF document or file package and place it as an invisible text layer on top of the original images. The result is saved in PDF format, which is the industry standard for storing and sharing documents. The resulting files can be searched and indexed, and the text can be selected and copied in the same way as if you selected and copied the original characters.

Bulk recognition

Aspose.OCR for Python via .NET allows you to recognize multiple files, regardless of their number and type, as simple as reading a single image. With a single API call, you can recognize multiple pages from an auto-feed scanner or extract vehicle license plates from automatic traffic cameras.

The results can be saved as searchable PDF document or spreadsheets, or returned as plain text, JSON or XML for further analysis.