This project is aim at taking input image which may be in a different orientation and align this image. This project was mentored and guided by - Rishiraj Acharya,Rajwrita,Ritesh.
- Align image of pages #2: In this task, I was responsible for Align image of pages that are in different orientations using keypoint matching and a homography matrix so that we can apply OCR to the texts in the page.
- Reading Image: First, we are going to take input using argparse and read it using opencv
- Image preprocessing: This step includes converting our images to grayscale for edge detection
- Edge Detection: Then we will carry out edge detection using canny edge detection.
- Finding Contours: First we will find contours of the image to get the cordinates of the edge of documents
- Draw Bounding Box: Then we draw bounding box around our document for visualization purpose
- Four Points Transformation: First we will apply four points transformation to align our document
- Gaussian Blur: Then we will apply gausian blur to give scanned effect on the document
- Python
- Opencv
- imutils
- numpy
- argparse
- Saving Resulting Output: Our script will save the the resulting output image as 'Scanned.png'
Here is the link to my PR: image_align_scan.py #6
- I believe this project not only helps in document alignment but also opens up a new world of possiblities for DSC-NSEC. We can use this project as an advanced feature for document scanning in apps and many other.
- I had an enriching and exciting experience in winter of code,2020, working on DocScanner under DSC_NSEC. I was glad to be mentored by Rishiraj Acharya,Rajwrita,Ritesh(thanks for going through my PRs!). I express my sincere thanks to all students who worked on DocScanner as part of WOC,2020.