Font detection

Aspose.OCR for Python via .NET can detect font information for each recognized text line. This is useful when you need to preserve visual characteristics of recognized content or analyze documents by typography.

Font detection is enabled through RecognitionSettings.detect_fonts. When the setting is enabled, each non-empty item in RecognitionResult.recognition_lines_result can contain a FontLineResult object in its font property.

Font detection is designed for recognized text lines. Enable line-level recognition output and use clear, sufficiently large text images for best results.

Font information

FontLineResult contains the following information:

Property	Description
line_index	Zero-based line index in the processed font batch.
size_px	Detected line height in pixels.
size_pt_estimate	Estimated font size in points.
style	Detected font style, for example regular, italic, or bold_italic.
style_confidence	Confidence score for the detected style.
font	Detected font family name.
font_confidence	Confidence score for the detected font family.

Detecting fonts

Set RecognitionSettings.detect_fonts to True, recognize the image, and read the font property of each recognized line.

import aspose.ocr

api = aspose.ocr.AsposeOcr()

input_data = aspose.ocr.OcrInput(aspose.ocr.InputType.SINGLE_IMAGE)
input_data.add("source.png")

settings = aspose.ocr.RecognitionSettings()
settings.detect_fonts = True
settings.detect_areas_mode = aspose.ocr.DetectAreasMode.LEAN
settings.language = aspose.ocr.Language.LATIN
settings.threads_count = 1

result = api.recognize(input_data, settings)[0]

for line in result.recognition_lines_result:
    if not line.text_in_line or not line.text_in_line.strip():
        continue

    font = line.font

    print(line.text_in_line)
    print(f"Font: {font.font} ({font.font_confidence:.0%})")
    print(f"Style: {font.style} ({font.style_confidence:.0%})")
    print(f"Size: {font.size_px}px / {font.size_pt_estimate}pt")

Processing multiple images

Font detection works with batches. Each recognition result contains line-level font data for the corresponding input image or page.

import aspose.ocr

api = aspose.ocr.AsposeOcr()

input_data = aspose.ocr.OcrInput(aspose.ocr.InputType.SINGLE_IMAGE)
input_data.add("arial.png")
input_data.add("calibri.png")

settings = aspose.ocr.RecognitionSettings()
settings.detect_fonts = True
settings.detect_areas_mode = aspose.ocr.DetectAreasMode.LEAN
settings.language = aspose.ocr.Language.LATIN
settings.threads_count = 1

for result in api.recognize(input_data, settings):
    for line in result.recognition_lines_result:
        if not line.text_in_line or not line.text_in_line.strip():
            continue
        print(f"{line.text_in_line}: {line.font.font}, {line.font.style}")