Detecting text
< Previous help topic Next help topic >
The product allows you
to detect text in image and video profiles.
In order to detect face, select image or video profile of interest, right click it, choose Analyze images context menu item and then Detect text. The analysis will start.
After the analysis the
filter Images with text under Analysis results
profile subnode in Case Explorer will be filled with the images
with text detected. The total number of images with text will be indicated in the
filter node:
No images with text detected
The analysis is designed to work with clear scanned text. It might not work with arbitrary text like road signs, subtitles, text on T-shirts etc. The example of image which will be processed well is below:
To improve the results, the tool does some operations like de-skew, resolution increase and others. Thus, even small and skewed images may be processed.
To the current state of art it cannot be guaranteed that all text is extracted correctly. The following obstacles may prevent successful text detection:
- Significantly skewed text
- Too low resolution (e.g. letters of 5 pixels high)
- Hand-written marks over printed text
To see, what text is detected for an image, you can inspect Analysis properties section in Item Properties window:
In order to successfully recognize a text, you must specify recognition language in Options (English by default). Multilingual recognition is not supported.
Please note, that for text extraction from video file, you should extract key frames first.
