5. Image Processing

Every page until now has handled pixel data only long enough to get it: capture a frame, configure a sensor, move bytes from the camera into RAM. Image processing turns the direction around – the pixels are the input, and the goal is whatever the application needs from them.

The image module is the toolkit. Every operation takes one or two Image objects and returns either a transformed Image – pixels filtered, thresholded, warped, or composited differently – or a list of result objects: blobs, lines, codes, matches, statistics. The two flavours compose. A typical pipeline cleans up the captured frame, extracts a set of result objects from it, turns those into measurements, and draws annotations back onto the frame for display.

Foundations

• 5.1. The Image object
• 5.2. Coordinates and regions
• 5.3. Pixel formats
• 5.4. Reading and writing pixels
• 5.5. Regions and masks

Drawing

• 5.6. Drawing shapes and text
• 5.7. Compositing images
• 5.8. Flood fill and detection glyphs

Pixel math

• 5.9. Arithmetic operations
• 5.10. Bitwise operations
• 5.11. Frame differencing

Thresholding

• 5.12. Binary thresholding

Filters and morphology

• 5.13. Linear and neighbourhood filters
• 5.14. Gaussian smoothing and edges
• 5.15. Morphological operations
• 5.16. Custom convolution kernels
• 5.17. A catalogue of standard kernels

Tonal and statistical analysis

• 5.18. Histograms and statistics
• 5.19. Tonal corrections
• 5.20. Regression and similarity

Geometric transforms

• 5.21. Scale, flip, and crop
• 5.22. Lens and perspective correction
• 5.23. Perspective correction
• 5.24. Polar transforms

Finding shapes

• 5.25. Finding blobs
• 5.26. Finding lines and segments
• 5.27. Finding circles and rectangles

Decoding 1D and 2D codes

• 5.28. QR codes and AprilTags
• 5.29. Barcodes and Data Matrix codes

Matching and similarity

• 5.30. Template matching
• 5.31. Displacement matching

Image I/O

• 5.32. Saving and compression
• 5.33. ImageIO streams

Wrap up

• 5.34. Wrap up