When using the Recognize Text tool with screenshots and other images within the Image Editor, you might select text and receive an error that states, "No text detected". This can happen when the text you are selecting is too small, or the selected section contains graphics, images or other non-text elements.
If you receive this error, try the following:
- When selecting text, avoid including images, graphics, or any other items that are not text. For example, in the following image, rather than selecting the entire image at once, select the areas outlined in red separately for better results.
- When dragging to select an area of text, include a margin of the background around your selection.
In addition, be sure to use high-quality images with the Text Recognition tool. The recommended resolution for Optical Character Recognition (OCR) is 300 dpi. To meet this resolution, capture screenshots at 1024 x 768 pixels or more. Text that is blurry, or of lower quality is more difficult for the tool to detect. If you are having trouble detecting text, you can potentially improve text recognition results by:
- Using a higher resolution image or a zoomed-in version of your original image.
- Avoiding images that are blurry or contain out-of-focus text.
- Using an image that has better contrast, such as black text against a white background.
- Regenerating your source image using a larger font size.
- Maintaining strong contrast by using dark text on a light background or white text on a dark background
- Avoid "visual noise" such as complex backgrounds, watermarks, and overlapping graphics.
- Ensuring that the text within your image is straight and horizontal. Skewed or tilted text lowers text recognition accuracy.