Dmytro Khudenko: Challenges of implementing task managers in the corporate an...Lviv Startup Club
More Related Content
Similar to Юрій Іванов Тема: Програмні лічильники однотипних рухомих об’єктів: алгоритми, реалізація і особливості застосування у системах відеоспостереження
Similar to Юрій Іванов Тема: Програмні лічильники однотипних рухомих об’єктів: алгоритми, реалізація і особливості застосування у системах відеоспостереження (20)
Invezz.com - Grow your wealth with trading signals
Юрій Іванов Тема: Програмні лічильники однотипних рухомих об’єктів: алгоритми, реалізація і особливості застосування у системах відеоспостереження
1. Software counters of the similar-
typed moving objects in
videosurveillance systems
algorithms, implementation and application features
Yurii IVANOV
ELSY Co. Ltd. Software engineer. PhD.
2. KEY TOPICS
• formation, transmission and storing principles of digital
videostreams. Basic components of videosurveillance systems;
• moving objects representation to further analysis in digital
videoframes areas of interest;
• videoframes analysis, composition of binary frames areas interest:
thresholding, smoothing, gradients, mathematical morphology
operations (with example on chicken carcasses at suspended
chain conveyors counting unit);
• moving objects tracking and trajectories estimations in
videoframes areas of interest (with example on bakery productions
at tunnel ovens counting unit);
• users interfaces, software modules settings monitoring in real
time. Counting results registration and displaying.
3. Computerized accounting and control
systems
Surveillance system scheme
Key capabilities:
- real-time videostreams monitoring and analyzing;
- videodata storing (continuous and by event);
- previewing stored videofiles and data analysis
results.
4. Multiscreen for real time streams monitoring
- previewing every channel in separated window;
- availability to manage every channel and quick preview;
- instant manual record for every channel;
- organization of multiscreen (split into different windows
configurations);
5. Streams recording management
- storing videodata at any local, remote storage disks and DVRs;
- cyclic rewriting of video archives;
- manage of free disks space;
- schedule for weekly storage planning;
- manual records and by events (e.g. motion, sound detection).
6. Previewing stored videodata
- availability to synchronous preview videoarchives from every
source local and remote;
- calendar that displays all of records within each day;
- extended timeline that matches time of manual and by events
recordings;
7. Structure of videoanalysis modules
Every videoanalysis module is implemented as separated dll-file
that expands the basic functionalities without making any changes
into the base surveillance system.
8. Chicken carcasses at suspended chain conveyors
counting unit
1. Select two regions of interest ( for
carcass and its holder).
2. Convert frame to HSV color space
and split it on components.
3. For each region of interest create
rotation matrix and use warpAffine
OpenCV library function to rotate it
within few degrees.
4. Use S component of HSV image to
represent a conveyor holder image.
5. Use V component of HSV image to
represent a chicken carcass image.
9. Chicken carcasses at suspended chain conveyors
counting unit
6. For conveyor holder image and for
carcasses image use two different
thresholds to create a binary masks.
7. Applying morphology operations to
remove some amount of small BLOBs
that may appear on a masks and fill a
holes that may appear within BLOBs
of interest.
8. For conveyor holder binary image
threshold by BLOBs height is used to
obtain rectangulars representation.
9. Applying area threshold for each
rectangular BLOB.
10. Chicken carcasses at suspended chain conveyors
counting unit
10. Merge two binary masks.
11. For each rectangular that
satisfying area threshold and for
corresponding carcass holder draw a
central line.
12. At carcass image binary mask
choose rectangular area.
13. Calculate an amount of
foreground pixels within each
rectangular area and compare it to
predefined area threshold.
14. If amount of foreground pixel is
larger than predefined area
threshold, draw a circle at the
bottom of each area.
15. Analyse each circle on crossing
predefined counting line.
14. Counting units results representation
The results from counting unit are stored on MySQL
database and can be represented with histograms. By default result
data are displayed for a current day. User can display data for any
time interval with histograms steps: by month, weeks, days, hours or
minutes and for any videochannel from which results were recorded.
Also previewing data module contains some functions that allows to
save a displayed data to XLS, HTML, CSV and image files