peakfinders2D.py 文件源码

def find_peaks_regionprops(z, min_sigma=4, max_sigma=5, threshold=1, 
                           min_size=50, return_props=False):
    """
    Finds peaks using regionprops.
    Uses the difference of two gaussian convolutions to separate signal from 
    background, and then uses the skimage.measure.regionprops function to find 
    connected islands (peaks). Small blobs can be rejected using `min_size`.

    Parameters
    ----------
    z : ndarray
        Array of image intensities.
    min_sigma : int, float
        Standard deviation for the minimum gaussian convolution
    max_sigma : int, float
        Standard deviation for the maximum gaussian convolution
    threshold : int, float
        Minimum difference in intensity
    min_size : int
        Minimum size in pixels of blob
    return_props : bool
        Return skimage.measure.regionprops

    Returns
    -------
    ndarray
        (n_peaks, 2)
        Array of peak coordinates.

    """
    from skimage import morphology, measure

    difference = ndi.gaussian_filter(z, min_sigma) - ndi.gaussian_filter(z, max_sigma)

    labels, numlabels = ndi.label(difference > threshold)
    labels = morphology.remove_small_objects(labels, min_size)

    props = measure.regionprops(labels, z)

    if return_props:
        return props
    else:
        peaks = np.array([prop.centroid for prop in props])
        return clean_peaks(peaks)