presentation sim opencv

Post on 12-Jul-2015

221 Views

Category:

Engineering

3 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Simulation with Computer Vision

Andres Fernandez

Outline

● Computer Vision– Algorithms

– Object Tracking Technique

● Robot Simulation

Computer Vision

● Hough Transform Circle

● Smallest-circle

Hough Transform Circle

● Detects circles

● Outputs the X and Y coordinates

● Outputs the radius

Smallest-circle

● Finds circles

● Outputs the X and Y coordinates

● Outputs the radius

● Finds the smallest circle out of all circles

Computer Vision Object Tracking

● Changing the environment to make calculations easier

● Detecting circles using Hough Transform Circle

● Using the output to feed into simulation

Changing the environment

● Environment color change from RGB to HSV● HSV stands for Hue Saturation Value● Hue – Color shown● Saturation – Color intensity● Value – Color brightness

Changing the environment

Source: http://en.wikipedia.org/wiki/File:Hsl-hsv_models.svg

RGB vs. HSV

Hough Transformer Circle

● Needs three parameters (x, y, r)●

● Changed the equation to find x and y as the center

(x−a)2+( y−b)2

=r2

x=a+R cos(θ) y=b+R sin(θ)

Hough Transformer Circle

● An 2-dim accumulation array keeps track of the number of intersections

● The pixel with highest count is considered to be center of the circle

● w = width, h = height

O(wh(δR))

δR=(MaximumRadius−MinimumRadius)

Hough Transformer Circle

Simulation

● A simple 2D animation of a box

● The box represents a robot which follows an object of interest

● The velocity is fixed

● The turning velocity can be changed in the GUI

Simulation

Threading

● Three threads total

● GUI – Main thread

● Image Processing – Child Thread

● Robot Simulation – Child Thread

Scheduling

● Image thread

Scheduling

● Simulation thread

Improvements

● Background Subtraction

● OpenCL (Multicore)

● More Benchmarking Analysis

● Extending Simulation (more variables)

● Real Time Operating System (RTOS)

Questions

References

● Alper Yilmaz, Omar Javed, and Mubarak Shah. 2006. Object tracking: A survey. ACM Comput. Surv. 38, 4, Article 13 (December 2006)

● Canny, John, "A Computational Approach to Edge Detection," Pattern Analysis and Machine Intelligence, IEEE Transactions on , vol.PAMI-8, no.6, pp.679,698, Nov. 1986

● H. K. Yuen, J. Princen, J. Illingworth, and J. Kittler. 1990. Comparative study of Hough transform methods for circle finding. Image Vision Comput. 8, 1 (February 1990), 71-77

● Jasmin Blanchette and Mark Summerfield , “C++ GUI Programming with Qt 4”, 2nd ed., Prentice Hall, 2008

Extra - Reasons

● Why use actual sensors to connect to the simulation?

● What framework to use to display the video and animation?

Extra - Technical Specifications

● Intel Core i7

● 6 Gigabytes of Memory

● OpenCV 2.4.8

● Webcam: 15 MP

● Webcam: 30 FPS

Extra – Edge Detection

● Edge Detection is used to simplify the image

● The Hough Transform Circle function uses Canny and Sobel to detect edges

● Sobel is a predecessor of Canny

● Canny and Sobel is a multi step process

Extra – Canny Edge Detection

● Color to Gray scale● Gaussian filter● Gradient Processing● Non-maximum suppression● Tracing edges● Hysteresis Thresholding

top related