Sony SNC-CS50P - Manual (Page 12)

12
In outdoor surveillance and monitoring applications,
surveillance cameras are usually attached to poles or
mounted on buildings. Depending on the installation site,
captured video might be displayed as shaky images
resulting from vibration caused by wind and other
environmental effects. The image stabilizer function
incorporated in the Sony SNC-RX550/RZ50/CS50 Series of
network cameras minimizes the effect caused by high-
and low-frequency vibration to provide stable images.
This function is especially useful for outdoor surveillance
and traffic-monitoring applications.
The image stabilizer mechanism works as follows: when
the image stabilizer function is activated, the camera
assigns a 5% border area in the image to compensate for
camera vibration (Fig. 1).
*1 An 8 x 8 matrix of calculation
blocks (reference areas in the image) are assigned as
calculation points for motion vectors (Fig. 2). The camera
stores 30 frames of this image data in its memory.
Movement of each calculation block is compared frame by
frame using what is called a Block Matching Algorithm
(BMA),
*2 and individual motion vectors are calculated
(Fig. 3). The system is such that individual motion vectors
associated with movement of objects or beings within the
image are disregarded. Individual motion vectors
associated with movement of the camera are then
processed to obtain an average resultant motion vector
for the entire image (Fig. 4). The camera then shifts the
image to the direction opposite that of the resultant
motion vector to correct for any camera movement (Fig.
5). Because the camera stores data from the past 30
frames, these motion vectors are continually updated and
corrected in such a manner that the resultant image is
smooth and natural (i.e. the correction process is performed
incrementally so as not to cause abrupt transitions in the
image).
The image stabilizer function is effective in environments
where the vibration frequency of the camera is
approximately 2 Hz. This function can be used when the
camera is operating in MPEG-4, JPEG, and dual-encoding
(MPEG-4/JPEG) modes as well as when “Intelligent Motion
Detection” or “Intelligent Object Detection” are active.
Image Stabilizer
5% (24 pixeles in VGA)
Calculation blocks
Motion vector generated by
bird movement
Image shift to the direction opposite
that of the resultant motion vector
Motion vector
Disregards motion vector
generated by bird movement
5% (32 pixeles in VGA)
Previous Frame Current Frame
Previous Frame Current Frame
Fig. 1 Border Area (5% of image)
Fig. 2 Calculation Blocks (8 x 8 matrix)
Fig. 3 Individual Motion Vectors
Fig. 4 Resultant Motion Vector for Camera Movement
Fig. 5 Image Shift to Compensate for Camera Movement
*1 Digital zoom is used to allow for compensation; therefore, the effective
viewing area is reduced. Depending on the settings, the camera might
operate at a lower frame rate due to the amount of processing
required by the camera.
*2 BMA is an algorithm for locating matching blocks in a sequence of
video frames for the purposes of motion compensation.