tracking_types.h File Reference

Value types shared by the motion detector and the tracker (pure C++20, natively tested). More...

#include <array>
#include <cstdint>

Go to the source code of this file.

Classes
struct	tracking::Box
	Inclusive cell rectangle. A default-constructed box is empty. More...
struct	tracking::Blob
	One connected region of foreground cells. More...
struct	tracking::DetectorConfig
	Motion detector tunables. Defaults are tuned for an indoor scene 1-3 m away. More...
struct	tracking::DetectionResult
	What the detector found in one frame. Blobs are sorted by area, largest first. More...
struct	tracking::TrackerConfig
	Tracker tunables (frame counts assume ~17 fps). More...
struct	tracking::TrackState
	The tracked target. Positions in cells, velocity in cells per frame. id 0 = no target. More...

Typedefs
using	tracking::GrayFrame = std::array<uint8_t, kCells>
	One 40x30 luma frame, row-major (index = y * kGridW + x).

Enumerations
enum class	TrackPhase : uint8_t { Idle = 0 , Tentative = 1 , Locked = 2 , Coasting = 3 }

Functions
const char *	tracking::to_string (TrackPhase phase)

Variables
constexpr int	tracking::kGridW = 40
constexpr int	tracking::kGridH = 30
constexpr int	tracking::kCells = kGridW * kGridH
constexpr int	tracking::kMaxBlobs = 16

Detailed Description

Value types shared by the motion detector and the tracker (pure C++20, natively tested).

Everything works on a 40x30 gray grid: the camera's QVGA JPEG decoded at 1/8 scale (one DC value per 8x8 block). Coordinates are in grid cells; multiply by 8 for QVGA pixels.