6. NumPy

The Image class covers the per-pixel work the camera already knows how to do – thresholds, blob finding, edge detection, and the rest of the image library. The numpy module covers everything else: ADC readings and other buffers of plain numbers, math that runs across a whole buffer at once, and per-pixel transforms the image library has not already built in.

Enter the ndarray, a single class that holds a packed block of equally typed numbers. Everything else in numpy is a math function that operates on an ndarray. Adding two arrays together, summing one of them, taking a sine of every element – each is a single library call that processes the whole buffer in one go, much faster than the equivalent Python for loop.

Concepts

• 6.1. Why arrays
• 6.2. The ndarray
• 6.3. Making arrays
• 6.4. Dtypes

Shape and indexing

• 6.5. Shape and strides
• 6.6. Indexing and slicing
• 6.7. Views and copies

Math

• 6.8. Operators
• 6.9. Universal functions
• 6.10. Broadcasting
• 6.11. Reductions
• 6.12. Selection and rearrangement

Linear algebra

• 6.13. Linear algebra

Signal processing

• 6.14. FFT
• 6.15. Filtering and spectrograms

Numerical tools

• 6.16. Curves and integration
• 6.17. Solvers and random numbers

Images

• 6.18. Images and ndarrays

Performance

• 6.19. Performance

Wrap up

• 6.20. Wrap up