How to recognize traffic sign when it is small in image sequence ?
Hi all, I am working on traffic sign recognition using template matching. I am able to find correct result when vehicle is near to traffic sign. But when it is away, means candidate is small in image, result is poor & false. So,please suggest how should i approach ( any algo, idea, concept etc..) to solve this problem. For reference please see cropped taffic sign from image sequence ( to make it easy ). How i can make to recognize correctly at/before image 1 or brfore image 2 ? Any suggestion, i am just 3 month old in computer vision. Thanks in advance.
All the robust commercial and open source systems out there use trained models for this. Template matching will never cover the possibilities that these techniques give you ... look at HOG+SVM, cascade classifiers, Random Forests, ...