[Various image analysis with plotly] Dynamic visualization with plotly [python, image]

Let's analyze the image easily with plotly

environment

python==3.8 plotly==4.10.0 scikit-image==0.17.2 requests==2.24.0 Pillow==7.2.0 matplotlib==3.3.2

Handling of common images

I will use pillow first

import matplotlib.pyplot as plt
from PIL import Image
import requests
import io

url = 'https://upload.wikimedia.org/wikipedia/commons/thumb/0/00/Crab_Nebula.jpg/240px-Crab_Nebula.jpg'
img = Image.open(io.BytesIO(requests.get(url).content))

plt.figure(figsize=(5,5))
plt.subplot(111)
plt.imshow(img)

I was able to display the image with the usual matplot Let's make it black and white

gray_img = img.convert('L')

plt.figure(figsize=(5,5))
plt.subplot(111)
plt.imshow(gray_img)

image object can be black and white using convert

Image processing using skimage and plotly

The nice thing about displaying images with plotly is that you can zoom in, etc.

import plotly.express as px
from skimage import io
img_sk = io.imread('https://upload.wikimedia.org/wikipedia/commons/thumb/0/00/Crab_Nebula.jpg/240px-Crab_Nebula.jpg')
fig = px.imshow(img_sk)
fig.show()

You can also handle images read with pillow

fig = px.imshow(gray_img, color_continuous_scale='gray')
fig.show()

Create contour lines in the image

You can add contour lines to the brightness of the image

Contour plot to color the value of Z Monitor the values from start to end Specify the interval of contour lines with size By making end and size the same, you can extract the part that has the specified value. In other words, it can be used for contour extraction, etc.

import plotly.graph_objects as go
fig = px.imshow(gray_img, color_continuous_scale='gray')
fig.add_trace(go.Contour(z=gray_img, showscale=True,
                         contours=dict(start=0, end=30, size=30,coloring='lines'),line_width=1))

fig.show()

import plotly.express as px
import plotly.graph_objects as go
from skimage import data
img_camera = data.camera()
fig = px.imshow(img_camera, color_continuous_scale='gray')

fig.add_trace(go.Contour(z=img_camera, showscale=False,
                         contours=dict(start=0, end=70, size=70, coloring='lines'),
                         line_width=2))
fig.show()

Create a color channel histogram from plotly

from plotly.subplots import make_subplots
from skimage import data

img_sk.shape

#(240, 240, 3)

plt.imshow(img_sk)


img_r = img_sk.copy()
img_r[:, :, 1] = 0
img_r[:, :, 2] = 0
plt.imshow(img_r)

img_g = img_sk.copy()
img_g[:, :, 0] = 0
img_g[:, :, 2] = 0
plt.imshow(img_g)

img_b = img_sk.copy()
img_b[:, :, 0] = 0
img_b[:, :, 1] = 0
plt.imshow(img_b)

Non-zero to hist

ff=go.Figure()
ff.add_trace(go.Histogram(x=img_r.flatten()[img_r.flatten()!=0],marker_color='red',name='red'))
ff.add_trace(go.Histogram(x=img_g.flatten()[img_g.flatten()!=0],marker_color='green',name='green'))
ff.add_trace(go.Histogram(x=img_b.flatten()[img_b.flatten()!=0],marker_color='blue',name='blue'))
ff.update_layout(barmode='overlay')
ff.update_traces(opacity=0.3)
ff.update_layout(height=400)

ff.show()

When expanding, it seems that 10 or less is the main, so check the distribution of 10 or more

ff=go.Figure()
ff.add_trace(go.Histogram(x=img_r.flatten()[img_r.flatten()>10],marker_color='red',name='red'))
ff.add_trace(go.Histogram(x=img_g.flatten()[img_g.flatten()>10],marker_color='green',name='green'))
ff.add_trace(go.Histogram(x=img_b.flatten()[img_b.flatten()>10],marker_color='blue',name='blue'))
ff.update_layout(barmode='overlay')
ff.update_traces(opacity=0.3)
ff.update_layout(height=400)

ff.show()

It looks like this when it is divided neatly

Confirmation for black and white is as follows

from skimage.color import rgb2gray

gray_img = rgb2gray(img_sk)


ff=go.Figure()
ff.add_trace(go.Histogram(x=gray_img.flatten()[gray_img.flatten()!=0],marker_color='pink',name='gray'))
ff.update_layout(barmode='overlay')
ff.update_traces(opacity=0.8)
ff.update_layout(height=400)

ff.show()

ff=go.Figure()
ff.add_trace(go.Histogram(x=gray_img.flatten()[gray_img.flatten()>0.1],marker_color='pink',name='gray'))
ff.update_layout(barmode='overlay')
ff.update_traces(opacity=0.8)
ff.update_layout(height=400)

ff.show()

By the way, how to display the image in the figure

Appeal to the entire surface of the graph

import plotly.express as px
fig = px.scatter(x=[0, 1, 2, 3, 4], y=[0, 1, 4, 9, 16])

fig.add_layout_image(
        dict(
            source="https://upload.wikimedia.org/wikipedia/commons/thumb/0/00/Crab_Nebula.jpg/240px-Crab_Nebula.jpg ",
            xref="x",
            yref="y",
            x=0,
            y=16,
            sizex=4,
            sizey=15,
            sizing="stretch")
)

fig.update_xaxes(title_text="picture No")
fig.update_yaxes(title_text="view par day", hoverformat=".3f")

fig.show()

Gently appeal on the back (below)

import plotly.express as px
fig = px.scatter(x=[0, 1, 2, 3, 4], y=[0, 1, 4, 9, 16])

fig.add_layout_image(
        dict(
            source="https://upload.wikimedia.org/wikipedia/commons/thumb/0/00/Crab_Nebula.jpg/240px-Crab_Nebula.jpg ",
            xref="x",
            yref="y",
            x=0,
            y=16,
            sizex=4,
            sizey=15,
            sizing="stretch",
            opacity=0.5,
            layer="below")
)

fig.update_xaxes(title_text="picture No")
fig.update_yaxes(title_text="view par day", hoverformat=".3f")

fig.show()

Make it smaller and add a stick or put text

import plotly.express as px
fig = px.scatter(x=[0, 1, 2, 3, 4], y=[0, 1, 4, 9, 16])

fig.add_layout_image(
        dict(
            source="https://upload.wikimedia.org/wikipedia/commons/thumb/0/00/Crab_Nebula.jpg/240px-Crab_Nebula.jpg ",
            xref="x",
            yref="y",
            x=1.68,
            y=16,
            sizex=3,
            sizey=3)
)

fig.update_layout(
    annotations=[
        dict(
            x=0.5,
            y=0.8,
            xref="paper",
            yref="paper",
            showarrow=True,
            arrowhead=0,
            opacity=0.5,
            ax=190,
            ay=100,
        )#,
#        dict(x=,y=,xref="paper",yref="paper",showarrow=True,arrowhead=0,opacity=0.5,ax=,ay=,)
    ]
)

fig.update_xaxes(title_text="picture No")
fig.update_yaxes(title_text="view par day", hoverformat=".3f")

fig.show()

The layout image of the image conforms to the position of the plot The x, y axis of the plot corresponds to the upper left position x, y of the image

The bar is 0,0 at the bottom left of the image Center is 0.5,0.5 ax and ay represent the length of the bar, and extend from the specified positions x and y to the vector-combined position of the bars extending in the x-axis direction and the y-axis direction.

that's all

Anything fun theory if you can move it for the time being