Image Matching (Template Matching)

Template Matching is a method in image processing for searching the location of a template or a target image.

Outline

• Template Matching
• Template Matching for Multi-Objects

Template Matching

- File name: Image_Matching.py 
- Input image: brain.jpg 
- Input target: target.jpg
- Command Line: python Image_Matching.py -i brain.jpg -t target.jpg

• Main Function: res = cv2.matchTemplate(src_img, target, method)

Method:
   * cv2.TM_CCOEFF            * cv2.TM_CCORR_NORMED
   * cv2.TM_CCOEFF_NORMED     * cv2.TM_SQDIFF
   * cv2.TM_CCORR             * cv2.TM_SQDIFF_NORMED

• The mathematic description for methods describe in OpenCV-Python Documentation

NOTE: use function cv2.minMaxLoc() to calculate the "rectangle" of the target.
  - min_val, max_val, min_loc, max_loc = cv2.minMaxLoc(res)
    * If method use "cv2.TM_SQDIFF", "cv2.TM_SQDIFF_NORMED", top_left = min_loc.
      Else, top_left = max_loc
    
  - cv2.rectangle(src_img, top_left, bottom_right, color, thickness)
    * top_left = min_loc or max_loc (depend on the method)
    * bottom_right = (top_left[0] + width, top_left[1] + height)

Template Matching for Multi-objects

- File name: Image_Matching.py 
- Input image: cell.jpg
- Input target: cell_target.jpg
- Command Line: python Image_Matching.py -i cell.jpg -t cell_target.jpg

Set a threshold for the return value by using numpy.where() for cv2.matchTemplate(), then 
draw all the region using cv2.rectangle():

* location = np.where( res >= threshold )

Useful link:

• Changing Colorspace

• Image Thresholding

Code

• Image Matching (Template Matching)
• Template Matching Implementation by Using "Normalized Cross Correlation"

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Reference & Acknowledgments

• OpenCV-Python Tutorial: https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_tutorials.html
• (minMaxLoc) https://docs.opencv.org/2.4/modules/core/doc/operations_on_arrays.html#void%20minMaxLoc(InputArray%20src,%20double*%20minVal,%20double*%20maxVal,%20Point*%20minLoc,%20Point*%20maxLoc,%20InputArray%20mask)
• (matchTemplate) https://docs.opencv.org/2.4/modules/imgproc/doc/object_detection.html
• (Drawing function) https://docs.opencv.org/2.4/modules/core/doc/drawing_functions.html