Added OMR test scanner using CV techniques

[Panchadip-128]

Introduction
OMR (Optical Mark Recognition) is a technique used to capture human-marked data from documents like multiple-choice exams, surveys, and ballots. An OMR scanner identifies marks made by users (typically filled bubbles or boxes) on a printed form, and then processes that data. This project aims to build an OMR scanner and grader that can read and grade multiple-choice test forms including handling of multiple marked bubbles , Non-filled bubbles, Visualization of correct and incorrect answers and displaying the percentage of correct answers at the bottom of the OMR sheet. It also includes reading the answer key from a CSV file, easier to input large answer keys.

Methodology
Image Preprocessing:
Image Acquisition: Capture or scan the OMR sheet (scanned or photographed).

Grayscale Conversion: Convert the image to grayscale to simplify further processing.

Thresholding/Binarization: Apply thresholding to convert the grayscale image into a binary image (black and white). This helps in clearly distinguishing the filled marks from the background.

Noise Reduction: Use filters (e.g., Gaussian or median filters) to remove noise from the image.

OMR Sheet Alignment: Perspective Transformation: Correct any skew or rotation in the scanned form using perspective transformation. Contour Detection: Find the contours of the OMR form, ensuring to process the area that contains the marks (typically a grid of answer choices). Bubble/Mark Detection: Grid Division: Divide the detected region of interest (ROI) into sections corresponding to the answer options (i.e., the grid where bubbles are located). Mark Identification: For each section, check whether a bubble is filled. This can be done by counting the number of black pixels in each bubble region and setting a threshold to decide if a bubble is marked (e.g., more than 60% of the area is filled). -Handling Multiple Marks: If the user marks more than one option for a question, it can be marked as incorrect based on predefined rules. Answer Key Comparison: Once the marked bubbles are detected, compare the selected answers with the correct answer key stored in the system. The given code reads the answer key from a CSV file using the pandas library. Each correct answer receives a point, and wrong or multiple answers get no points.

[github-actions]

Thank you for submitting your pull request! 🙌 We'll review it as soon as possible. If there are any specific instructions or feedback regarding your PR, we'll provide them here. Thanks again for your contribution! 😊

[Panchadip-128]

@Niketkumardheeryan Sorry to intervene, but pls review once as time is less

[Niketkumardheeryan]

@Panchadip-128 run each cell of your code ,