Module trackpal.features

Module for track feature descriptors.

Functions

def angles_from_displacements(disp_xy)

Compute the angle in radians from track displacements

Args

disp_xy : numpy.array

displacements [n x 2]

Returns

numpy.array

angle in radioans [n x 1]

Classes

class ConfinementRatio (coords, frame)

Compute confinement ration as net distance divided by total distance travelled $$\frac{\|x_n-x_1\|}{\sum_{i=1}^n \|x_i\|}$$

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class DirectionalChange (coords, frame)

Compute statistics of the angle between two successive positions

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class DirectionalChangeCount (coords, frame)

Count of angles exceeding 20, 45, or 90 degree angle turns

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class Features (coords, frame)

Factory class for the creating of TrackFeature

Static methods

def list(verbose=False)

List names of all available features

Methods

def get(self, name)

Retrieve track feature by name

class GyrationTensor (coords, frame)

Compute gyration tensor by fitting an ellipse and extraction several measures

• axis lengths
• axis ratio
• radius $R^2$
• coherence

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class MSDParabola (coords, frame)

Fits a parabola to mean squared displacement curve of track and returns fit parameters and $R^2$

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj, clip=0.9, min_trj_len=3)

class MeanStraightLineSpeed (coords, frame)

Compute mean straight line speed as net distance divided by total time $$\frac{\|x_n-x_1\|}{\sum_{i=1}^n \|\Delta t_i\|}$$

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class PartitionFeature (coords, frame)

Compute features based on spatio-temporal clustering with DBSCAN.

Track is partitioned by partition_dbscann and features of stationary (dwelling) and moving tracklets are computed.

For the stationary part:

• partition_dwell_state_count
• partition_dwell_state_total_duration
• partition_dwell_state_relative_duration
• partition_dwell_between_state_duration_mean
• partition_dwell_between_state_duration_max
• partition_dwell_between_state_length_mean
• partition_dwell_between_state_length_max

For the moving tracklets

• partition_moving_speed_stats_mean__mean
• partition_moving_speed_stats_std__mean
• partition_moving_speed_stats_min__mean
• partition_moving_speed_stats_max__mean
• partition_moving_gyration_tensor_minor_axis_length__mean
• partition_moving_gyration_tensor_major_axis_length__mean
• partition_moving_gyration_tensor_axis_ratio__mean
• partition_moving_gyration_tensor_radius__mean
• partition_moving_gyration_tensor_coherence__mean
• partition_moving_directional_change_mean__mean

Ancestors

• TrackFeature

Class variables

var name

Static methods

def dwell_centroid_idx(labels)

Track indicies of dwell sub tracks

Args

labels : np.array

labels from dbscann

Returns

list[np.array]

List of sub-track indicies

def moving_sub_tracks(trj, labels, coords, frame, minimum_size=3)

Returns the moving tracklets of given track and labels

Args

trj : pandas.DataFrame

The track

labels : np.array

labels from dbscann

coords : list[str]

xy identifiers

frame : str

frame identifier

Returns

list[pandas.DataFrame]

List of moving subtracks

def partition_dbscann(p, t, eps, t_scale=1.0)

Partition track by spatio-temporal clustering using DBSCAN

Args

p : numpy.array

xy location

t : numpy.array

times

eps : float

DBSCAN epsilon

t_scale : float, optional

[description]. Defaults to 1.0.

Returns

numpy.array

DBSCAN label assingment

Methods

def compute(self, trj, eps=6, t_scale=1.0)

class RDP_simplification (coords, frame)

Applies Ramer–Douglas–Peucker algorithm to simplify the track for given epsilon and extracts:

• ratio of positions in simplified and original track
• mean and std directional change

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj, min_trj_len=3, eps=1)

class SpeedStats (coords, frame)

Compute several statistics of the instantaneous speed

Satistics

• mean
• std
• min
• max

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj, min_trj_len=2)

class TrackDiameter (coords, frame)

Statistics of the track diameter

• mean
• std
• max

of all pairwise distances of all positions in the track

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class TrackDuration (coords, frame)

Duration of track

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class TrackFeature (coords, frame)

Base class for all track features

Subclasses

• ConfinementRatio
• DirectionalChange
• DirectionalChangeCount
• GyrationTensor
• MSDParabola
• MeanStraightLineSpeed
• PartitionFeature
• RDP_simplification
• SpeedStats
• TrackDiameter
• TrackDuration
• TrackLength
• VACstats
• VelocityAverage

class TrackLength (coords, frame)

Length of track

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)

class VACstats (coords, frame)

Computes velocity autocorrelation and returns statistics

Satistics

• mean
• std
• min
• max

of the curve for delays $> 0$

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj, min_trj_len=3)

class VelocityAverage (coords, frame)

Average velocity

Ancestors

• TrackFeature

Class variables

var name

Methods

def compute(self, trj)