PvaRadarCFARParamsRec#

Defined in public/src/operator/include/OpRadarCFAR.h

struct PvaRadarCFARParamsRec#

Configuration parameters for radar CFAR target detection operator.

This structure contains all the necessary parameters to configure the Constant False Alarm Rate (CFAR) algorithm for radar target detection. The parameters control the detection sensitivity, algorithm type, and processing window dimensions for optimal target detection performance.

The CFAR algorithm uses training cells to estimate the noise floor and guard cells to protect the cell under test (CUT) from target energy leakage, ensuring accurate threshold calculation.

CFAR Cell Arrangement: N = numTrain K = numGuard

* | T1 | T2 | T3 |......| TN | G1 | ... | GK | CUT | G1 | ... | GK | T1 | T2 | T3 |......| TN |
* |<-------- Trailing ------>|<--- Guard --->|     |<--- Guard --->|<-------- Leading ------->|
*

Public Members

int32_t numHorTrain#

Number of horizontal training cells for CFAR algorithm (range: 0 < numHorGuard < numHorTrain <= (Width-1)/2) Constraint: 0 <= numHorGuard < numHorTrain < 256 Constraint: 1 <= numHorTrain + numHorGuard < 256 Training cells are used to estimate the noise floor in the horizontal direction.

int32_t numHorGuard#

Number of horizontal guard cells for CFAR algorithm (range: 0 <= numHorGuard < numHorTrain) Constraint: 0 <= numHorGuard < numHorTrain < 256 Guard cells protect the cell under test from target energy leakage in horizontal direction.

int32_t numVerTrain#

Number of vertical training cells for CFAR algorithm (range: 0 < numVerGuard < numVerTrain <= (Height-1)/2) Constraint: 0 <= numVerGuard < numVerTrain < 256 Constraint: 1 <= numVerTrain + numVerGuard < 256 Training cells are used to estimate the noise floor in the vertical direction.

int32_t numVerGuard#

Number of vertical guard cells for CFAR algorithm (range: 0 <= numVerGuard < numVerTrain) Constraint: 0 <= numVerGuard < numVerTrain < 256 Guard cells protect the cell under test from target energy leakage in vertical direction.

float horizontalThresholdFactor#

CFAR horizontal threshold multiplier Controls detection sensitivity - higher values increase detection threshold noiseEstimate is the noise estimate in the horizontal direction detectionThreshold = noiseEstimate * thresholdFactor.

float verticalThresholdFactor#

CFAR vertical threshold multiplier Controls detection sensitivity - higher values increase detection threshold detectionThreshold = noiseEstimate * thresholdFactor noiseEstimate is the noise estimate in the vertical direction.

bool isHorizontalCyclicPadding#

Controls padding mode for horizontal dimension during CFAR processing.

When enabled (true):

  • Uses cyclic padding, Uses the training cells from the opposite edge for boundary training cells.

  • Recommended when horizontal data exhibits periodic/cyclic characteristics or when background noise and clutter properties are assumed to be consistent across edges.

When disabled (false):

  • Uses available neighbouring pixels for noise estimation at boundaries.

  • Training cells beyond image boundaries are not included in noise estimation.

  • Recommended when the signal represents an event that is known to start and end within the collected data window, with a genuinely zero-level background outside that window.

bool isVerticalCyclicPadding#

Controls padding mode for vertical dimension during CFAR processing.

When enabled (true):

  • Uses cyclic padding, Uses the training cells from the opposite edge for boundary training cells.

  • Recommended when vertical data exhibits periodic/cyclic characteristics or when background noise and clutter properties are assumed to be consistent across edges.

When disabled (false):

  • Uses available neighbouring pixels for noise estimation at boundaries.

  • Training cells beyond image boundaries are not included in noise estimation.

  • Recommended when the signal represents an event that is known to start and end within the collected data window, with a genuinely zero-level background outside that window.

bool enablePeakGrouping#

Enables peak grouping algorithm for target detection refinement When enabled, the algorithm filters detections to retain only local maxima by comparing each detection against its immediate horizontal and vertical neighbors.