Simple Python library for fitting a (rectangular) grid of shapes in an image to a set of points.
Note that the current beta version still lacks a number of features.
This module requires Python >= 3.8 and the following packages:
numpyscipyopencv-pythonmatplotlib
Install with pip
$ pip install git+https://github.com/nelsond/gridfitFirst, find the dominant angle of a rectangular grid in an image (uses the standard deviation of the Radon transformation).
import numpy as np
from gridfit.rect import find_dominant_angle
image = np.load('data.npy')
angle = find_dominant_angle(image)After finding the dominant angle of the rectangular grid, we can fit Gaussian-shaped blobs to the one-dimensional projection of the data. This allows us to extract the coordinates of all points on the rectangular grid.
import matplotlib.pyplot as plt
from gridfit.rect import fit_grid
grid = fit_grid(image, angle, min_rel_height=0.2)
plt.imshow(image)
for col in grid:
for row in col:
plt.plot(row[1], row[0], 'ro', mfc='none', ms=10)Once the grid has been determined, we can use a set of ROIs to analyze the data.
from gridfit.roi import CircularROI, ROIDataset
rois = []
for point in grid.reshape(-1, 2):
roi = CircularROI(point, 15)
rois.append(roi)
ds = ROIDataset(image, rois)
ds.plot(show_center=False, imshow_kwargs=dict(cmap=plt.cm.Greys_r))
# determine sum across each ROI on the grid
summed = ds.sum().reshape(*grid.shape[:2], -1)
# determine centroid in each ROI (first moment)
centroids = ds.centroid(absolute=True).reshape(*grid.shape[:2], -1)
# determine rms size in each ROI (square root of the second moment)
rms_size = ds.rms_size().reshape(*grid.shape[:2], -1)Install requirements for development environment
$ pip install .[dev]Run tests
$ pytest tests/Generate coverage report
$ pytest --cov=gridfit --cov-report html tests/
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent