Kalman filter and Hungarian algorithm for multiple object
tracking
The vision.KalmanFilter object
is designed for object tracking. You can use it to predict an
object's future location, to reduce noise in the detected location,
or to help associate multiple objects with their corresponding
tracks. The configureKalmanFilter function
helps you to set up the Kalman filter object.
The assignDetectionsToTrack function
assigns detections to tracks in the context of multiple object
tracking using the James Munkres' variant of the Hungarian
assignment algorithm. The function also determines which tracks are
missing, and which detections should begin a new track.
Image and video annotation for detected or tracked objects
The insertObjectAnnotation function
inserts labels and corresponding circles or rectangles into an
image or video to easily display tracked objects. You can use it
with either a grayscale or true color image input.
Kanade-Lucas-Tomasi (KLT) point tracker
The vision.PointTracker object
tracks a set of points using the Kanade-Lucas-Tomasi (KLT), feature
tracking algorithm. You can use the point tracker for video
stabilization, camera motion estimation, and object tracking.
HOG-based people detector
The vision.PeopleDetector object
detects people in an input image using the Histogram of Oriented
Gradient (HOG) features and a trained Support Vector Machine (SVM)
classifier. The object detects unoccluded people in an upright
position.
Video file reader support for H.264 codec (MPEG-4) on Windows
7
This release adds H.264 codec (MPEG-4) video formats for Windows 7
operating systems.
Show matched features display
The showMatchedFeatures function
displays corresponding feature points. It displays a falsecolor
overlay of two images with a color-coded plot of the corresponding
points connected by a line.
Matching methods added for match features
function
This release enhances the matchFeatures function
for applications in computing the fundamental matrix, stereo
vision, registration, and object detection. It provides three
different matching methods: simple threshold match, unique matches,
and unambiguous matches.
Compatibility Considerations
The new implementation of matchFeatures uses
different default value for the method parameter.
If you need the same results as those produced by the previous
implementation, set the Methodparameter
with syntax:
matchFeatures(FEATURES1, FEATURES2, 'Method', 'NearestNeighbor_old',
...).
Kalman filter for tracking tutorial
The Kalman filter is a popular tool for object tracking.
The Using
Kalman Filter for Object Trackingexample helps you to
understand how to setup and use the vision.KalmanFilter object
and theconfigureKalmanFilter function
to track objects.
Motion-based multiple object tracking example
The Motion-Based
Multiple Object Tracking example shows you how
to perform automatic detection and motion-based tracking of moving
objects in a video from a stationary camera.
Face detection and tracking examples
The Face
Detection and Tracking example shows you how
to automatically detect and a track a face.
The Face
Detection and Tracking Using the KLT
Algorithm example uses the Kanade-Lucas-Tomasi
(KLT) algorithm and shows you how to automatically detect a face
and track it using a set of feature points.
Stereo image rectification example
This release enhances the Stereo Image Rectification example. It
uses SURF feature detection with theestimateFundamentalMatrix, estimateUncalibratedRectification,
and detectSURFFeaturesfunctions
to compute the rectification of two uncalibrated images, where the
camera intrinsics are unknown.
System object tunable parameter support in code generation
You can change tunable properties in user-defined System objects at
any time, regardless of whether the object is locked. For System
objects predefined in the software, the object must be locked. In
previous releases, you could tune System object properties only for
a limited number of predefined System objects in generated
code.
save and load for
System objects
You can use the save method
to save System objects to a MAT file. If the object is locked, its
state information is saved, also. You can recall and use those
saved objects with the load method.
You can also create your own save and load methods
for a System object you create. To do so, use
the saveObjectImpl and loadObjectImpl,
respectively, in your class definition file.
Save and restore SimState not
supported for System objects
The Save and Restore Simulation State as SimState option is no
longer supported for any System object in a MATLAB Function block.
This option was removed because it prevented parameter tunability
for System objects, which is important in code generation.
Compatibility Considerations
If you need to save and restore simulation states, you may be able
to use a corresponding Simulink block, instead of a System
object.