Assume that the speed limit at a place is captured and is provided as an input to a system which checks if the speed of vehicle is above the accepted level or not. For checking, the number in the speed limit image has to be extracted.

Can anyone suggest a simple method to perform number segmentation and identification in the above case? Also if K-means algorithm can be used?

  • $\begingroup$ Welcome to DSP SE! Why not to start marking answers to your questions as accepted, so that everyone can benefit from that and you can increase your reputation? $\endgroup$ – jojek May 26 '15 at 9:43

As with all ocr, you need to segment each individual character. Assuming this step is already done (and maybe some post processing to One easy way is to take the number and reduce the resolution. Similar to a lcd character display. enter image description here

You can then do a simple correlation between the reduced resolution number from the image, and a template. Since the numbers on street signs are printed and very consistent a simple template match could work well.

your template should contain the numbers 0-9 in a font that closely matches those used on your road signs.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.