Revision history [back]

Hi I am trying to make a modified OMR election/survey system. Where a user can mark circle his choice will be marked accordingly to the text next to the circle. So far I have used blob detection and I can extract the coordinates via keypoints. I am thinking of using these coordinates and evaluate if it is within a specific rectangle to evaluate the user choice. I tried using findcontours to do rectangle identification but I am having no results since it also identifies the text as contours and some other lines as individual contours. I am following this tutorial for the contours. I am also trying to filter the rects by area to isolate only the needed rects, but I am not sure how to do that. Here is my sample image

Hi I am trying to make a modified OMR election/survey system. Where a user can mark circle his choice will be marked accordingly to the text next to the circle. So far I have used blob detection and I can extract the coordinates via keypoints. I am thinking of using these coordinates and evaluate if it is within a specific rectangle to evaluate the user choice. I tried using findcontours to do rectangle identification but I am having no results since it also identifies the text as contours and some other lines as individual contours. I am following this tutorial for the contours. I am also trying to filter the rects by area to isolate only the needed rects, but I am not sure how to do that. Here is my sample image

Hi I am trying to make a modified OMR election/survey system. Where a user can mark circle his choice will be marked accordingly to the text next to the circle. So far I have used blob detection and I can extract the coordinates via keypoints. I am thinking of using these coordinates and evaluate if it is within a specific rectangle to evaluate the user choice. I tried using findcontours to do rectangle identification but I am having no results since it also identifies the text as contours and some other lines as individual contours. I am following this tutorial for the contours. I am also trying to filter the rects by area to isolate only the needed rects, but I am not sure how to do that. Here is my sample image

So I changed my approach. Basically, I made the circle marks red. so I can pre-process the image and use findcontours to detect only the circle marks. I am planning to use the coordinates from SimpleBlobDetector to check if the points are inside the circle contours. Will this approach work? Processing time-wise will another approach be better? Thanks.