Ask Your Question

Digit and road sign recognition

asked 2012-11-17 18:21:49 -0500

Szczepan gravatar image

updated 2018-12-10 04:03:21 -0500

Hello. My purpose is to make a self driving car, witch recognize digits(speed limit) and road sign(not complex ie left or right arrow). Can I do it with open CV? I have 1 month to do that and I don't want to waste time to read about that library that I can't use to make my project.(I want to program in C++ using raspberry pi and A4WD robot)

edit retag flag offensive close merge delete

3 answers

Sort by ยป oldest newest most voted

answered 2012-11-19 01:51:23 -0500

SR gravatar image


From the website:

Usage ranges from interactive art, to mines inspection, stitching maps on the web or through advanced robotics.

edit flag offensive delete link more

answered 2012-11-19 02:14:53 -0500

updated 2012-11-19 11:23:28 -0500

You can use machine learning (MLL) .It is a set of classes and functions for statistical classification, regression, and clustering of data. For recognize digits & signs you have to Create a database for sample then select a method for feature extraction and select a classifier for train of features. The proposed method for feature extraction use contours of shapes & HU moment and MLP_BP for classifier.

edit flag offensive delete link more

answered 2012-12-01 19:15:22 -0500

Szczepan gravatar image

I'm getting image from camera, use GaussianBlur(srcg,srcg,Size(5,5),0,0) and then Canny(srcg,srcg,low_tresh_slider,high_tresh_slider,3). I'm trying to use findContours(srcg,conturs,hierarchy,CV_RETR_TREE,CV_CHAIN_APPROX_SIMPLE,Point(0,0)) but I error like these

Assertion failed (k == STD_VECTOR_MAT) in release, file /build/buildd/opencv-2.3.1/modules/core/src/matrix.cpp, line 1364 terminate called after throwing an instance of 'cv::Exception' what(): /build/buildd/opencv-2.3.1/modules/core/src/matrix.cpp:1364: error: (-215) k == STD_VECTOR_MAT in function release

definicion of Mat srcg(src.size(),CV_8UC1); ( I read about findConturs function and it takes a single channel as a parameter). I am going in good direction with digit and sign recognition or should I use something different? I can make my own road signs so maybe it will be good idea to detect squares or circles (shape of sign) and don't process all image but just ROI defined by detected square or circle.

edit flag offensive delete link more


(1) Look at to find a potential approach. (2) If you have quesions on a particular topic open a new question but please do not mix answers, questions and unrelated follow-up questions within a single thread. It makes it a mess and unreadable.

SR gravatar imageSR ( 2012-12-03 21:39:56 -0500 )edit
Login/Signup to Answer

Question Tools


Asked: 2012-11-17 18:21:49 -0500

Seen: 3,403 times

Last updated: Dec 01 '12