We Keep Coding

how to maintain the surround view cameras (4 camera images) position in the point cloud

I am trying to generate 3D point cloud from 4 RGB-D images. I am able to do that with Open3D but I am unable to maintain the position of the images. You can find the camera_parameters.json here.
import open3d as o3d
import cv2
import os
import numpy as np
import matplotlib.pyplot as plt
import json
def load_json(path):
with open(path) as f:
return json.load(f)
def parse_camera_params(camera):
param = {}
param['camera'] = camera['camera']
param['depth'] = {}
param['depth']['fx'] = camera['K_depth'][0][0]
param['depth']['fy'] = camera['K_depth'][1][1]
param['depth']['cx'] = camera['K_depth'][0][2]
param['depth']['cy'] = camera['K_depth'][1][2]
param['depth']['K'] = camera['K_depth']
# ignore distCoeffs_depth's 5th (1000) and 6th (0) element
# since they are strange
param['depth']['distCoeffs'] = np.array(camera['distCoeffs_depth'][:5])
param['depth_width'] = camera['depth_width']
param['depth_height'] = camera['depth_height']
param['color'] = {}
param['color']['fx'] = camera['K_color'][0][0]
param['color']['fy'] = camera['K_color'][1][1]
param['color']['cx'] = camera['K_color'][0][2]
param['color']['cy'] = camera['K_color'][1][2]
param['color']['K'] = camera['K_color']
# ignore distCoeffs_color's 5th (1000) and 6th (0) element
# since they are strange
param['color']['distCoeffs'] = np.array(camera['distCoeffs_color'][:5])
param['color_width'] = camera['color_width']
param['color_height'] = camera['color_height']
# world to depth
w2d_T = np.array(camera['M_world2sensor'])
param['w2d_R'] = w2d_T[0:3, 0:3]
param['w2d_t'] = w2d_T[0:3, 3]
param['w2d_T'] = camera['M_world2sensor']
d2w_T = np.linalg.inv(w2d_T)
param['d2w_R'] = d2w_T[0:3, 0:3]
param['d2w_t'] = d2w_T[0:3, 3]
param['d2w_T'] = d2w_T
return param
if __name__ == '__main__':
data_dir = "data/"
camera_params = load_json(os.path.join(data_dir,
'camera_parameters.json'))
SVC_NUM = 4
pcd_combined = o3d.geometry.PointCloud()
for i in range(SVC_NUM):
param = parse_camera_params(camera_params['sensors'][i])
color = cv2.imread(os.path.join(data_dir, 'color_{:05d}.png'.format(i)))
color = cv2.cvtColor(color, cv2.COLOR_BGR2RGB)
depth = cv2.imread(os.path.join(data_dir, 'depth_{:05d}.png'.format(i)), -1)
# depth = depth * 0.001 # factor to scale the depth image from carla
o3d_color = o3d.geometry.Image(color)
o3d_depth = o3d.geometry.Image(depth)
rgbd_image = o3d.geometry.RGBDImage.create_from_tum_format(o3d_color, o3d_depth, False)
h, w = depth.shape
dfx, dfy, dcx, dcy = param['depth']['fx'], param['depth']['fy'], param['depth']['cx'], param['depth']['cy']
intrinsic = o3d.camera.PinholeCameraIntrinsic(w, h, dfx,dfy, dcx, dcy)
intrinsic.intrinsic_matrix = param['depth']['K']
cam = o3d.camera.PinholeCameraParameters()
cam.intrinsic = intrinsic
cam.extrinsic = np.array(param['w2d_T'])
pcd = o3d.geometry.PointCloud.create_from_rgbd_image(rgbd_image, cam.intrinsic, cam.extrinsic)
o3d.io.write_point_cloud("svc_{:05d}_v13.pcd".format(i), pcd)
pcd_combined += pcd
o3d.io.write_point_cloud("svc_global_v13.pcd", pcd_combined)
With the above code I am getting output of svc_global_v13.pcd like below
As you can see, all the images are projected into center. As indicated in the json file, I would like the images to be positioned as left, right, front and rear in the 3D point cloud.
May I know what is it I am missing here?

How can I make a PlotDataItem with unique IDs per distinct line segment?

I also posted in the pyqtgraph forum here.
My overall goal is to have several clickable regions overlaid on an image, and if the plot boundary of any region is clicked I get a signal with the ID of that region. Something like this:
If I use only one PlotDataItem with nan-separated curves then each boundary sends the same signal. However, using a separate PlotDataItem for each boundary makes the application extremely sluggish.
I ended up subclassing ScatterPlotItem and rewriting the pointsAt function, which does what I want. The problem now is I can't figure out the appropriate way to change the ScatterPlotItem's boundingRect. Am I on the right approach? Is there a better way of doing this?
import numpy as np
import pyqtgraph as pg
from pyqtgraph.Qt import QtCore, QtGui
class CustScatter(pg.ScatterPlotItem):
def pointsAt(self, pos: QtCore.QPointF):
"""
The default implementation only checks a square around each spot. However, this is not
precise enough for my needs. It also triggers when clicking *inside* the spot boundary,
which I don't want.
"""
pts = []
for spot in self.points(): # type: pg.SpotItem
symb = QtGui.QPainterPath(spot.symbol())
symb.translate(spot.pos())
stroker = QtGui.QPainterPathStroker()
mousePath = stroker.createStroke(symb)
# Only trigger when clicking a boundary, not the inside of the shape
if mousePath.contains(pos):
pts.append(spot)
return pts[::-1]
"""Make some sample data"""
tri = np.array([[0,2.3,0,1,4,5,0], [0,4,4,8,8,3,0]]).T
tris = []
xyLocs = []
datas = []
for ii in np.arange(0, 16, 5):
curTri = tri + ii
tris.append(curTri)
xyLocs.append(curTri.min(0))
datas.append(ii)
def ptsClicked(item, pts):
print(f'ID {pts[0].data()} Clicked!')
"""Logic for making spot shapes from a list of (x,y) vertices"""
def makeSymbol(verts: np.ndarray):
outSymbol = QtGui.QPainterPath()
symPath = pg.arrayToQPath(*verts.T)
outSymbol.addPath(symPath)
# From pyqtgraph.examples for plotting text
br = outSymbol.boundingRect()
tr = QtGui.QTransform()
tr.translate(-br.x(), -br.y())
outSymbol = tr.map(outSymbol)
return outSymbol
app = pg.mkQApp()
pg.setConfigOption('background', 'w')
symbs = []
for xyLoc, tri in zip(xyLocs, tris):
symbs.append(makeSymbol(tri))
"""Create the scatterplot"""
xyLocs = np.vstack(xyLocs)
tri2 = pg.PlotDataItem()
scat = CustScatter(*xyLocs.T, symbol=symbs, data=datas, connect='finite',
pxMode=False, brush=None, pen=pg.mkPen(width=5), size=1)
scat.sigClicked.connect(ptsClicked)
# Now each 'point' is one of the triangles, hopefully
"""Construct GUI window"""
w = pg.PlotWindow()
w.plotItem.addItem(scat)
plt: pg.PlotItem = w.plotItem
plt.showGrid(True, True, 1)
w.show()
app.exec()

Solved! It turns out unless you specify otherwise, the boundingRect of each symbol in the dataset is assumed to be 1 and that the spot size is the limiting factor. After overriding measureSpotSizes as well my solution works:
import numpy as np
import pyqtgraph as pg
from pyqtgraph.Qt import QtCore, QtGui
class CustScatter(pg.ScatterPlotItem):
def pointsAt(self, pos: QtCore.QPointF):
"""
The default implementation only checks a square around each spot. However, this is not
precise enough for my needs. It also triggers when clicking *inside* the spot boundary,
which I don't want.
"""
pts = []
for spot in self.points(): # type: pg.SpotItem
symb = QtGui.QPainterPath(spot.symbol())
symb.translate(spot.pos())
stroker = QtGui.QPainterPathStroker()
mousePath = stroker.createStroke(symb)
# Only trigger when clicking a boundary, not the inside of the shape
if mousePath.contains(pos):
pts.append(spot)
return pts[::-1]
def measureSpotSizes(self, dataSet):
"""
Override the method so that it takes symbol size into account
"""
for rec in dataSet:
## keep track of the maximum spot size and pixel size
symbol, size, pen, brush = self.getSpotOpts(rec)
br = symbol.boundingRect()
size = max(br.width(), br.height())*2
width = 0
pxWidth = 0
if self.opts['pxMode']:
pxWidth = size + pen.widthF()
else:
width = size
if pen.isCosmetic():
pxWidth += pen.widthF()
else:
width += pen.widthF()
self._maxSpotWidth = max(self._maxSpotWidth, width)
self._maxSpotPxWidth = max(self._maxSpotPxWidth, pxWidth)
self.bounds = [None, None]
"""Make some sample data"""
tri = np.array([[0,2.3,0,1,4,5,0], [0,4,4,8,8,3,0]]).T
tris = []
xyLocs = []
datas = []
for ii in np.arange(0, 16, 5):
curTri = tri + ii
tris.append(curTri)
xyLocs.append(curTri.min(0))
datas.append(ii)
def ptsClicked(item, pts):
print(f'ID {pts[0].data()} Clicked!')
"""Logic for making spot shapes from a list of (x,y) vertices"""
def makeSymbol(verts: np.ndarray):
plotVerts = verts - verts.min(0, keepdims=True)
symPath = pg.arrayToQPath(*plotVerts.T)
return symPath
app = pg.mkQApp()
pg.setConfigOption('background', 'd')
symbs = []
for xyLoc, tri in zip(xyLocs, tris):
symbs.append(makeSymbol(tri))
"""Create the scatterplot"""
xyLocs = np.vstack(xyLocs)
tri2 = pg.PlotDataItem()
scat = CustScatter(*xyLocs.T, symbol=symbs, data=datas, connect='finite',
pxMode=False, brush=None, pen=pg.mkPen(width=5), size=1)
scat.sigClicked.connect(ptsClicked)
# Now each 'point' is one of the triangles, hopefully
"""Construct GUI window"""
w = pg.PlotWindow()
w.plotItem.addItem(scat)
plt: pg.PlotItem = w.plotItem
plt.showGrid(True, True, 1)
w.show()
app.exec()

Finding cells of a .stl file with negative mean curvature using VTK in python

I have a .stl file and i'm trying to find the coordinates of cells with negative mean curvature using VTK and python. I have wrote these codes which are working fine to change the colors of cells based on their mean curvature but what i'm willing to achieve is coordinates of exact cells and triangles with specific mean curvature, e.g. 3d coordinates of cells with most negative mean curvature.
Here are the codes:
import vtk
def gaussian_curve(fileNameSTL):
colors = vtk.vtkNamedColors()
reader = vtk.vtkSTLReader()
reader.SetFileName(fileNameSTL)
reader.Update()
curveGauss = vtk.vtkCurvatures()
curveGauss.SetInputConnection(reader.GetOutputPort())
curveGauss.SetCurvatureTypeToGaussian() # SetCurvatureTypeToMean() works better in the case of kidney.
ctf = vtk.vtkColorTransferFunction()
ctf.SetColorSpaceToDiverging()
p1 = [0.0] + list(colors.GetColor3d("MidnightBlue"))
p2 = [1.0] + list(colors.GetColor3d("DarkRed"))
ctf.AddRGBPoint(*p1)
ctf.AddRGBPoint(*p2)
cc = list()
for i in range(256):
cc.append(ctf.GetColor(float(i) / 255.0))
lut = vtk.vtkLookupTable()
lut.SetNumberOfColors(256)
for i, item in enumerate(cc):
lut.SetTableValue(i, item[0], item[1], item[2], 1.0)
lut.SetRange(0, 0) # In the case of kidney, the (0, 0) worked better.
lut.Build()
cmapper = vtk.vtkPolyDataMapper()
cmapper.SetInputConnection(curveGauss.GetOutputPort())
cmapper.SetLookupTable(lut)
cmapper.SetUseLookupTableScalarRange(1)
cActor = vtk.vtkActor()
cActor.SetMapper(cmapper)
return cActor
def render_scene(my_actor_list):
renderer = vtk.vtkRenderer()
for arg in my_actor_list:
renderer.AddActor(arg)
namedColors = vtk.vtkNamedColors()
renderer.SetBackground(namedColors.GetColor3d("SlateGray"))
window = vtk.vtkRenderWindow()
window.SetWindowName("Render Window")
window.AddRenderer(renderer)
interactor = vtk.vtkRenderWindowInteractor()
interactor.SetRenderWindow(window)
# Visualize
window.Render()
interactor.Start()
if __name__ == '__main__':
fileName = "400_tri.stl"
my_list = list()
my_list.append(gaussian_curve(fileName))
render_scene(my_list)
This code produce red cells for positive mean curvature and blue for negative ones.
I need the result(coordinates of cells) in the form of arrays or something like that.
I would appreciate any suggestion and help on this problem.

A possible solution with vtkplotter:
from vtkplotter import *
torus1 = Torus().addCurvatureScalars().addScalarBar()
print("list of scalars:", torus1.scalars())
torus2 = torus1.clone().addScalarBar()
torus2.threshold("Gauss_Curvature", vmin=-15, vmax=0)
show(torus1, torus2, N=2) # plot on 2 separate renderers
print("vertex coordinates:", len(torus2.coordinates()))
print("cell centers :", len(torus2.cellCenters()))
check out the resulting screenshot here
Additional example here.
Hope this helps.

So i found the answer from kitware weblog, here is the code that works fine using vtk.numpy_interface and vtk.util.numpy_support, but still it does not produce the normals_array and i don't know why??
import vtk
from vtk.numpy_interface import dataset_adapter as dsa
from vtk.util.numpy_support import vtk_to_numpy
def curvature_to_numpy(fileNameSTL, curve_type='Mean'):
colors = vtk.vtkNamedColors()
reader = vtk.vtkSTLReader()
reader.SetFileName(fileNameSTL)
reader.Update()
# Defining the curvature type.
curve = vtk.vtkCurvatures()
curve.SetInputConnection(reader.GetOutputPort())
if curve_type == "Mean":
curve.SetCurvatureTypeToMean()
else:
curve.SetCurvatureTypeToGaussian()
curve.Update()
# Applying color lookup table.
ctf = vtk.vtkColorTransferFunction()
ctf.SetColorSpaceToDiverging()
p1 = [0.0] + list(colors.GetColor3d("MidnightBlue"))
p2 = [1.0] + list(colors.GetColor3d("DarkOrange"))
ctf.AddRGBPoint(*p1)
ctf.AddRGBPoint(*p2)
cc = list()
for i in range(256):
cc.append(ctf.GetColor(float(i) / 255.0))
lut = vtk.vtkLookupTable()
lut.SetNumberOfColors(256)
for i, item in enumerate(cc):
lut.SetTableValue(i, item[0], item[1], item[2], 1.0)
lut.SetRange(0, 0) # In the case of kidney, the (0, 0) worked better.
lut.Build()
# Creating Mappers and Actors.
mapper = vtk.vtkPolyDataMapper()
mapper.SetInputConnection(curve.GetOutputPort())
mapper.SetLookupTable(lut)
mapper.SetUseLookupTableScalarRange(1)
actor = vtk.vtkActor()
actor.SetMapper(mapper)
# Scalar values to numpy array. (Curvature).
dataObject = dsa.WrapDataObject(curve.GetOutput())
normals_array = dataObject.PointData['Normals'] # Output array.
curvature_array = dataObject.PointData['Mean_Curvature'] # output array.
# Node values to numpy array.
nodes = curve.GetOutput().GetPoints().GetData()
nodes_array = vtk_to_numpy(nodes)
# Creating a report file (.vtk file).
writer = vtk.vtkPolyDataWriter()
writer.SetFileName('vtk_file_generic.vtk')
writer.SetInputConnection(curve.GetOutputPort())
writer.Write()
# EDIT:
# Creating the point normal array using vtkPolyDataNormals().
normals = vtk.vtkPolyDataNormals()
normals.SetInputConnection(reader.GetOutputPort()) # Here "curve" could be replaced by "reader".
normals.ComputePointNormalsOn()
normals.SplittingOff()
normals.Update()
dataNormals = dsa.WrapDataObject(normals.GetOutput())
normals_array = dataNormals.PointData["Normals"]
return actor, normals_array, curvature_array, nodes_array
def render_scene(my_actor_list):
renderer = vtk.vtkRenderer()
for arg in my_actor_list:
renderer.AddActor(arg)
namedColors = vtk.vtkNamedColors()
renderer.SetBackground(namedColors.GetColor3d("SlateGray"))
window = vtk.vtkRenderWindow()
window.SetWindowName("Render Window")
window.AddRenderer(renderer)
interactor = vtk.vtkRenderWindowInteractor()
interactor.SetRenderWindow(window)
# Visualize
window.Render()
interactor.Start()
if __name__ == '__main__':
filename = "400_tri.stl"
my_list = list()
my_actor, my_normals, my_curve, my_nodes = curvature_to_numpy(filename, curve_type="Mean")
my_list.append(my_actor)
render_scene(my_list) # Visualization.
print(my_nodes) # Data points.
print(my_normals) # Normal vectors.
print(my_curve) # Mean curvatures.

Draw network and grouped vertices of the same community or partition

I need view (drawn or plot) the communities structure in networks
I have this:
import igraph
from random import randint
def _plot(g, membership=None):
layout = g.layout("kk")
visual_style = {}
visual_style["edge_color"] = "gray"
visual_style["vertex_size"] = 30
visual_style["layout"] = layout
visual_style["bbox"] = (1024, 768)
visual_style["margin"] = 40
for vertex in g.vs():
vertex["label"] = vertex.index
if membership is not None:
colors = []
for i in range(0, max(membership)+1):
colors.append('%06X' % randint(0, 0xFFFFFF))
for vertex in g.vs():
vertex["color"] = str('#') + colors[membership[vertex.index]]
visual_style["vertex_color"] = g.vs["color"]
igraph.plot(g, **visual_style)
if __name__ == "__main__":
karate = igraph.Nexus.get("karate")
cl = karate.community_fastgreedy()
membership = cl.as_clustering().membership
_plot(karate, membership)
But the vertices are spread. In another networks this result is very worse.
I want the vertices are grouped by color in a similar region.
E.g:

Based on #gabor-csardi answer, I made this code:
import igraph
from random import randint
def _plot(g, membership=None):
if membership is not None:
gcopy = g.copy()
edges = []
edges_colors = []
for edge in g.es():
if membership[edge.tuple[0]] != membership[edge.tuple[1]]:
edges.append(edge)
edges_colors.append("gray")
else:
edges_colors.append("black")
gcopy.delete_edges(edges)
layout = gcopy.layout("kk")
g.es["color"] = edges_colors
else:
layout = g.layout("kk")
g.es["color"] = "gray"
visual_style = {}
visual_style["vertex_label_dist"] = 0
visual_style["vertex_shape"] = "circle"
visual_style["edge_color"] = g.es["color"]
# visual_style["bbox"] = (4000, 2500)
visual_style["vertex_size"] = 30
visual_style["layout"] = layout
visual_style["bbox"] = (1024, 768)
visual_style["margin"] = 40
visual_style["edge_label"] = g.es["weight"]
for vertex in g.vs():
vertex["label"] = vertex.index
if membership is not None:
colors = []
for i in range(0, max(membership)+1):
colors.append('%06X' % randint(0, 0xFFFFFF))
for vertex in g.vs():
vertex["color"] = str('#') + colors[membership[vertex.index]]
visual_style["vertex_color"] = g.vs["color"]
igraph.plot(g, **visual_style)
if __name__ == "__main__":
g = igraph.Nexus.get("karate")
cl = g.community_fastgreedy()
membership = cl.as_clustering().membership
_plot(g, membership)
Results:

Remove the edges across multiple communities, calculate the layout without these edges, and then use it for the original graph.

To group the vertices of a community together and highlight them you should use 'mark_groups=True'. See http://igraph.org/python/doc/igraph.clustering-pysrc.html#VertexClustering.plot

Python+Chaco+Traits - rendering bug: unexpected fills of line plot of large data?

Using the minimal example below, the line plot of a large (some 110k points) plot I get (with python 2.7, numpy 1.5.1, chaco/enable/traits 4.3.0) is this:
However, that is bizarre, because it is a line plot, and there shouldn't be any filled areas in there? Especially since the data is sawtooth-ish signal? It's as if there is a line at y~=37XX, above which there is color filling?! But sure enough, if I zoom into an area, I get the rendering I expect - without the unexpected fill:
Is this a bug - or is there something I'm doing wrong? I tried to use use_downsampling, but it makes no difference...
The test code:
import numpy as np
import numpy.random as npr
from pprint import pprint
from traits.api import HasTraits, Instance
from chaco.api import Plot, ArrayPlotData, VPlotContainer
from traitsui.api import View, Item
from enable.component_editor import ComponentEditor
from chaco.tools.api import PanTool, BetterSelectingZoom
tlen = 112607
alr = npr.randint(0, 4000, tlen)
tx = np.arange(0.0, 30.0-0.00001, 30.0/tlen)
ty = np.arange(0, tlen, 1) % 10000 + alr
pprint(len(ty))
class ChacoTest(HasTraits):
container = Instance(VPlotContainer)
traits_view = View(
Item('container', editor=ComponentEditor(), show_label=False),
width=800, height=500, resizable=True,
title="Chaco Test"
)
def __init__(self):
super(ChacoTest, self).__init__()
pprint(ty)
self.plotdata = ArrayPlotData(x = tx, y = ty)
self.plotobj = Plot(self.plotdata)
self.plotA = self.plotobj.plot(("x", "y"), type="line", color=(0,0.99,0), spacing=0, padding=0, alpha=0.7, use_downsampling=True)
self.container = VPlotContainer(self.plotobj, spacing=5, padding=5, bgcolor="lightgray")
#~ container.add(plot)
self.plotobj.tools.append(PanTool(self.plotobj))
self.plotobj.overlays.append(BetterSelectingZoom(self.plotobj))
if __name__ == "__main__":
ChacoTest().configure_traits()

I am able to reproduce the error and talking with John Wiggins (maintainer of Enable), it is a bug in kiva (which chaco uses to paint on the screen):
https://github.com/enthought/enable
The good news is that this is a bug in one of the kiva backend that you can use. So to go around the issue, you can run your script choosing a different backend:
ETS_TOOLKIT=qt4.qpainter python <NAME OF YOUR SCRIPT>
if you use qpainter or quartz, the plot looks (on my machine) as expected. If you choose qt4.image (the Agg backend), you will reproduce the issue. Unfortunately, the Agg backend is the default one. To change that, you can set the ETS_TOOLKIT environment variable to that value:
export ETS_TOOLKIT=qt4.qpainter
The bad news is that fixing this isn't going to be an easy task. Please feel free to report the bug in github (again https://github.com/enthought/enable) if you want to be involved in this. If you don't, I will log it in the next couple of days. Thanks for reporting it!

Just a note - I found this:
[Enthought-Dev] is chaco faster than matplotlib
I recall reading somewhere that you are expected to implement the
_downsample method because the optimal algorithm depends on the type
of data you're collecting.
And as I couldn't find any examples with _downsample implementation other than decimated_plot.py referred in that post, which isn't standalone - I tried and built a standalone example, included below.
The example basically has messed up drag and zoom, (plot disappears if you go out of range, or stretches upon a drag move) - and it starts zoomed in; but it is possible to zoom it out in the range shown in the OP - and then it displays the exact same plot rendering problem. So downsampling isn't the solution per se, so this is likely a bug?
import numpy as np
import numpy.random as npr
from pprint import pprint
from traits.api import HasTraits, Instance
from chaco.api import Plot, ArrayPlotData, VPlotContainer
from traitsui.api import View, Item
from enable.component_editor import ComponentEditor
from chaco.tools.api import PanTool, BetterSelectingZoom
#
from chaco.api import BaseXYPlot, LinearMapper, AbstractPlotData
from enable.api import black_color_trait, LineStyle
from traits.api import Float, Enum, Int, Str, Trait, Event, Property, Array, cached_property, Bool, Dict
from chaco.abstract_mapper import AbstractMapper
from chaco.abstract_data_source import AbstractDataSource
from chaco.array_data_source import ArrayDataSource
from chaco.data_range_1d import DataRange1D
tlen = 112607
alr = npr.randint(0, 4000, tlen)
tx = np.arange(0.0, 30.0-0.00001, 30.0/tlen)
ty = np.arange(0, tlen, 1) % 10000 + alr
pprint(len(ty))
class ChacoTest(HasTraits):
container = Instance(VPlotContainer)
traits_view = View(
Item('container', editor=ComponentEditor(), show_label=False),
width=800, height=500, resizable=True,
title="Chaco Test"
)
downsampling_cutoff = Int(4)
def __init__(self):
super(ChacoTest, self).__init__()
pprint(ty)
self.plotdata = ArrayPlotData(x = tx, y = ty)
self.plotobj = TimeSeriesPlot(self.plotdata)
self.plotobj.setplotranges("x", "y")
self.container = VPlotContainer(self.plotobj, spacing=5, padding=5, bgcolor="lightgray")
self.plotobj.tools.append(PanTool(self.plotobj))
self.plotobj.overlays.append(BetterSelectingZoom(self.plotobj))
# decimate from:
# https://bitbucket.org/mjrosen/neurobehavior/raw/097ef3719d1263a8b303d29c31ab71b6e792ab04/cns/widgets/views/decimated_plot.py
def decimate(data, screen_width, downsampling_cutoff=4, mode='extremes'):
data_width = data.shape[-1]
downsample = np.floor((data_width/screen_width)/4.)
if downsample > downsampling_cutoff:
return globals()['decimate_'+mode](data, downsample)
else:
return data
def decimate_extremes(data, downsample):
last_dim = data.ndim
offset = data.shape[-1] % downsample
if data.ndim == 2:
shape = (len(data), -1, downsample)
else:
shape = (-1, downsample)
data = data[..., offset:].reshape(shape).copy()
data_min = data.min(last_dim)
data_max = data.max(last_dim)
return data_min, data_max
def decimate_mean(data, downsample):
offset = len(data) % downsample
if data.ndim == 2:
shape = (-1, downsample, data.shape[-1])
else:
shape = (-1, downsample)
data = data[offset:].reshape(shape).copy()
return data.mean(1)
# based on class from decimated_plot.py, also
# neurobehavior/cns/chaco_exts/timeseries_plot.py ;
# + some other code from chaco
class TimeSeriesPlot(BaseXYPlot):
color = black_color_trait
line_width = Float(1.0)
line_style = LineStyle
reference = Enum('most_recent', 'trigger')
traits_view = View("color#", "line_width")
downsampling_cutoff = Int(100)
signal_trait = "updated"
decimate_mode = Str('extremes')
ch_index = Trait(None, Int, None)
# Mapping of data names from self.data to their respective datasources.
datasources = Dict(Str, Instance(AbstractDataSource))
index_mapper = Instance(AbstractMapper)
value_mapper = Instance(AbstractMapper)
def __init__(self, data=None, **kwargs):
super(TimeSeriesPlot, self).__init__(**kwargs)
self._index_mapper_changed(None, self.index_mapper)
self.setplotdata(data)
self._plot_ui_info = None
return
def setplotdata(self, data):
if data is not None:
if isinstance(data, AbstractPlotData):
self.data = data
elif type(data) in (ndarray, tuple, list):
self.data = ArrayPlotData(data)
else:
raise ValueError, "Don't know how to create PlotData for data" \
"of type " + str(type(data))
def setplotranges(self, index_name, value_name):
self.index_name = index_name
self.value_name = value_name
index = self._get_or_create_datasource(index_name)
value = self._get_or_create_datasource(value_name)
if not(self.index_mapper):
imap = LinearMapper()#(range=self.index_range)
self.index_mapper = imap
if not(self.value_mapper):
vmap = LinearMapper()#(range=self.value_range)
self.value_mapper = vmap
if not(self.index_range): self.index_range = DataRange1D() # calls index_mapper
if not(self.value_range): self.value_range = DataRange1D()
self.index_range.add(index) # calls index_mapper!
self.value_range.add(value)
# now do it (right?):
self.index_mapper = LinearMapper(range=self.index_range)
self.value_mapper = LinearMapper(range=self.value_range)
def _get_or_create_datasource(self, name):
if name not in self.datasources:
data = self.data.get_data(name)
if type(data) in (list, tuple):
data = array(data)
if isinstance(data, np.ndarray):
if len(data.shape) == 1:
ds = ArrayDataSource(data, sort_order="none")
elif len(data.shape) == 2:
ds = ImageData(data=data, value_depth=1)
elif len(data.shape) == 3:
if data.shape[2] in (3,4):
ds = ImageData(data=data, value_depth=int(data.shape[2]))
else:
raise ValueError("Unhandled array shape in creating new plot: " \
+ str(data.shape))
elif isinstance(data, AbstractDataSource):
ds = data
else:
raise ValueError("Couldn't create datasource for data of type " + \
str(type(data)))
self.datasources[name] = ds
return self.datasources[name]
def get_screen_points(self):
self._gather_points()
return self._downsample()
def _data_changed(self):
self.invalidate_draw()
self._cache_valid = False
self._screen_cache_valid = False
self.request_redraw()
def _gather_points(self):
if not self._cache_valid:
range = self.index_mapper.range
#if self.reference == 'most_recent':
# values, t_lb, t_ub = self.get_recent_range(range.low, range.high)
#else:
# values, t_lb, t_ub = self.get_range(range.low, range.high, -1)
values, t_lb, t_ub = self.data[self.value_name][range.low:range.high], range.low, range.high
#if self.ch_index is None:
# self._cached_data = values
#else:
# #self._cached_data = values[:,self.ch_index]
self._cached_data = values
self._cached_data_bounds = t_lb, t_ub
self._cache_valid = True
self._screen_cache_valid = False
def _downsample(self):
if not self._screen_cache_valid:
val_pts = self._cached_data
screen_min, screen_max = self.index_mapper.screen_bounds
screen_width = screen_max-screen_min
values = decimate(val_pts, screen_width, self.downsampling_cutoff,
self.decimate_mode)
if type(values) == type(()):
n = len(values[0])
s_val_min = self.value_mapper.map_screen(values[0])
s_val_max = self.value_mapper.map_screen(values[1])
self._cached_screen_data = s_val_min, s_val_max
else:
s_val_pts = self.value_mapper.map_screen(values)
self._cached_screen_data = s_val_pts
n = len(values)
t = np.linspace(*self._cached_data_bounds, num=n)
t_screen = self.index_mapper.map_screen(t)
self._cached_screen_index = t_screen
self._screen_cache_valid = True
return [self._cached_screen_index, self._cached_screen_data]
def _render(self, gc, points):
idx, val = points
if len(idx) == 0:
return
gc.save_state()
gc.set_antialias(True)
gc.clip_to_rect(self.x, self.y, self.width, self.height)
gc.set_stroke_color(self.color_)
gc.set_line_width(self.line_width)
#gc.set_line_width(5)
gc.begin_path()
#if len(val) == 2:
if type(val) == type(()):
starts = np.column_stack((idx, val[0]))
ends = np.column_stack((idx, val[1]))
gc.line_set(starts, ends)
else:
gc.lines(np.column_stack((idx, val)))
gc.stroke_path()
self._draw_default_axes(gc)
gc.restore_state()
if __name__ == "__main__":
ChacoTest().configure_traits()