VR-Assignment-Rutul 0.1.4

Coin Detection and Panorama Stitching

Coin Detection & Panorama Stitching

Part 1: Coin Detection

Overview

This module processes an input image to detect and extract individual coins using image processing techniques.

Steps Involved

1. Preprocessing (preProcessingImage)

   • Converts the image to grayscale for simplified processing.
   • Resizes the image to maintain consistency.
   • Applies Gaussian blur and adaptive thresholding to enhance edge detection.

2. Edge Detection (edgeDetection)

   • Identifies contours in the thresholded image.
   • Analyzes shape properties such as circularity and area to detect coin-like structures.
   • Draws detected edges on the original image.

3. Region-Based Segmentation (segmentCoins)

   • Generates a mask using detected contours.
   • Isolates the coin regions from the original image.

4. Extracting Individual Coins (extractEachCoin)

   • Determines the minimum enclosing circle for each detected coin.
   • Uses bitwise operations to isolate and crop each coin.

5. Counting Coins (countCoins)

   • Computes the total number of detected coins from the segmented results.

How to Run

1. Install dependencies:

   pip install -r requirements.txt
   

2. Navigate to the part1 directory:

   cd part1
   

3. Run the detection script:

   python coinDetection.py <input_image_path> <output_dir>
   

4. Example Run:

   python coinDetection.py coins/0.jpg output/
   

Output Files

• Edges_on_image.jpg – Image with detected coin edges outlined.
• coin_segmented.jpg – Image with extracted coin regions.
• coinX.jpg – Individual cropped images of each detected coin.

---

Part 2: Panorama Stitching

Overview

This module stitches multiple images together to create a seamless panorama using feature detection and homography estimation.

Steps Involved

1. Feature Detection & Extraction (siftDetectDescriptor)

   • Detects key points and extracts feature descriptors using the SIFT algorithm.

2. Keypoint Matching (interestPointMacher)

   • Matches keypoints between image pairs using BFMatcher and Lowe’s ratio test.

3. Homography Estimation (interestPointMacher)

   • Computes the transformation matrix (homography) for image alignment.
   • Utilizes RANSAC to filter outliers and refine accuracy.

4. Image Warping & Blending (stichImages)

   • Aligns images using the computed transformation.
   • Blends overlapping regions to create a seamless transition.

5. Cropping Unwanted Regions (cropBlackRegion)

   • Removes black borders caused by perspective transformation.
   • Extracts only the meaningful content of the stitched image.

How to Run

1. Install dependencies:

   pip install -r requirements.txt
   

2. Navigate to the part2 directory:

   cd part2
   

3. Run the panorama stitching script:

   python panorama.py <input_directory> <output_directory>
   

4. Example Run:

   python panorama.py input1/ output/
   

Output Files

• stitched_image_X.jpg – Visualization of matched keypoints between consecutive images.
• panorama.jpg – The final stitched panorama image.

---

Notes

• Ensure images for stitching have sufficient overlap for feature matching.
• The quality of results depends on image alignment and lighting conditions.