The MTV-2D Algorithm is specifically designed for 2D stereoscopic MTV, leveraging laser grid lines with a signal-to-noise ratio greater than 2 for optimal performance. This algorithm uses a three-staged cross-correlation optimization process to determine the correspondence between the source and target images.
- High Signal-to-Noise Ratio: Designed for laser grid lines with a signal-to-noise ratio > 2.
- Hough Line Transformation: Identifies grid patterns in the source image.
- Three-Stage Cross-Correlation: Ensures robust matching between source and target images.
The algorithm is design for the specific geometry of image generated by molecular tagging velocimetry using grid optical components. Below is an example of a source image with grid lines detected:
- Description: Example image from MTV.
-
Stage 1: Detect grid lines in the source image using Hough line transformation as the base template.
- Visualization:
- Visualization:
-
Stage 2: Perform initial coarse cross correspondence of image by mapping the rhombus region from target image with the source image.
- Visualization:
- Visualization:
-
Stage 3: Perform template matchign across two image to determine plausible region where corresponding intersection for source image exist in the target image.
- Visualization:
- Visualization:
-
Stage 4: Perform sub-pixel detection using template matching using the template generated from stage 1 with the determined search region in stage 3.
- Visualization:
- Visualization:
The final output includes a fully matched grid pattern between the source and target images, as shown below:
- Description: This image demonstrates the final matched grid, illustrating the algorithm's effectiveness in accurately determining correspondences.
- Dependencies: Python 3.8+, OpenCV, NumPy, SciPy
- Installation:
pip install -r requirements.txt