marimo notebook showing PIV image enhancement filter

enhance_piv_image_nb.py

import marimo

__generated_with = "0.19.6"
app = marimo.App(width="medium", auto_download=["ipynb"])

with app.setup:
    import cv2
    import numpy as np
    import matplotlib.pyplot as plt
    import marimo as mo


@app.function
def enhance_piv_particles(image, kernel_size=5, cap_percentile=95):
    # 1. White Top-Hat Transform
    # kernel_size should be slightly larger than your typical particle diameter (e.g., 3, 5, or 7)
    kernel = cv2.getStructuringElement(
        cv2.MORPH_ELLIPSE, (kernel_size, kernel_size)
    )
    tophat = cv2.morphologyEx(image, cv2.MORPH_TOPHAT, kernel)

    # 2. Local Contrast Enhancement (Optional but helpful for low intensity)
    # CLAHE spreads the intensity of small objects without global blooming
    clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8, 8))
    enhanced = clahe.apply(tophat)

    # 3. Intensity Capping
    # Prevents large bright regions from dominating the cross-correlation signal
    cap_value = np.percentile(enhanced, cap_percentile)
    enhanced_capped = np.clip(enhanced, 0, cap_value).astype(np.uint8)

    # Normalize back to 0-255 for visibility/processing
    final = cv2.normalize(enhanced_capped, None, 0, 255, cv2.NORM_MINMAX)

    return final


@app.cell
def _():
    # Usage:
    img = cv2.imread("/home/user/Desktop/C001H001S0001000001.png", 0)
    processed_img = enhance_piv_particles(img, kernel_size=5)
    return img, processed_img


@app.cell
def _(img, processed_img):
    _fig, (_ax_orig, _ax_proc) = plt.subplots(1, 2, figsize=(10, 5))

    _ax_orig.imshow(img, cmap="gray")
    _ax_orig.set_title("Original Image")
    _ax_orig.axis("off")

    _ax_proc.imshow(processed_img, cmap="gray")
    _ax_proc.set_title("Processed Image")
    _ax_proc.axis("off")

    plt.tight_layout()
    mo.mpl.interactive(_fig)
    return


if __name__ == "__main__":
    app.run()

molab.md