Q&A

What is the pyplot.imshow() method?

By AlgoInsight Team2026-06-07arrayspython

Overview

imshow ("image show") maps array values to colors and draws them as a pixel grid. A 2-D array becomes a heatmap colored by a colormap; a 3-D array (height × width × 3 or × 4) is shown as a real color image. It's ubiquitous in data science and image processing for visualizing matrices, model outputs, correlation heatmaps, and pictures.

graph LR
    A[2-D array → heatmap via colormap] --> C["plt.imshow(data)"]
    B[3-D array H×W×3 → RGB image] --> C
    C --> D[Rendered pixel grid]
    D --> E["plt.colorbar() adds a scale"]
    style C fill:#1e1b4b,color:#fff

Displaying a 2-D Array (heatmap)

import matplotlib.pyplot as plt
import numpy as np

data = np.random.rand(10, 10)        # 10x10 of values in [0,1]
plt.imshow(data, cmap="viridis")     # map values → colors
plt.colorbar()                        # show the value→color scale
plt.title("Random Heatmap")
plt.show()

Each array cell becomes a pixel; the cmap translates its value to a color. colorbar() adds the legend mapping colors back to values.

Displaying an Actual Image (3-D array)

import matplotlib.image as mpimg
img = mpimg.imread("photo.png")      # shape (H, W, 3) or (H, W, 4)
plt.imshow(img)                       # shown as a true color image
plt.axis("off")                       # hide axis ticks for photos
plt.show()

For 3-channel (RGB) or 4-channel (RGBA) arrays, imshow interprets the last axis as color and no colormap is used.

Key Parameters

plt.imshow(matrix, cmap="hot", vmin=0, vmax=1, interpolation="nearest")

The origin Gotcha

By default imshow puts row 0 at the top (origin="upper"), matching how images are stored. For mathematical plots where you want the y-axis increasing upward, use origin="lower". Mixing these up flips your visualization vertically - a frequent confusion.

imshow vs Related Functions

• imshow - regular grid / image; values mapped to pixels.
• pcolormesh / pcolor - grids with non-uniform coordinates (specify x/y edges).
• matshow - convenience wrapper around imshow tuned for matrices.
• contourf - filled contours rather than discrete pixels.

Practical Uses

• Heatmaps - correlation matrices, confusion matrices, attention maps.
• Image display - show photos, processed images, masks.
• Model visualization - feature maps, segmentation outputs.
• 2-D function visualization - render f(x, y) over a grid.

Why This Matters in Interviews

Data-science/visualization interviews use imshow to test plotting fluency. Key points: it renders a 2-D array as a colormapped heatmap or a 3-D array as an RGB image; cmap + colorbar() make values interpretable; vmin/vmax control the scale; and the origin default (row 0 at top) is a classic gotcha. Knowing interpolation="nearest" for honest pixel display (vs smoothing) and the matrix-oriented matshow shows depth.

Common Pitfalls

• Vertical flips from misunderstanding origin (upper vs lower).
• Forgetting colorbar(), leaving heatmap colors uninterpretable.
• Default interpolation smoothing data and implying detail that isn't there (use nearest).
• Passing a float image outside [0,1] or wrong dtype, producing clipped/odd colors.
• Forgetting plt.show() (or the inline backend) so nothing displays.

References

• Matplotlib imshow: https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.imshow.html
• Matplotlib image tutorial: https://matplotlib.org/stable/tutorials/introductory/images.html

More answers