# Project AutoVESTS

**Posted:**December 25, 2010

**Filed under:**Artificial intelligence, Image Processing, signal processing |

**Tags:**car, detection, final project, image, nokia 5800, opencv, opensift, opensurf, parallel port, recognition, traffic sign, webcam 1 Comment

Autonomous Vehicle System using Traffic Sign Recognition or Auto VESTS.

This is my final year project at university. The concept of the project is – A car which will move and turn different directions without any human support. So, this car can drive autonomously and can detect and recognize traffic sign. Now I want to include some advantages of this project which might distinguish our project from others.

- One of the major advantage I’d like to add is cost. Our system is very much cost effective and materials which we have used was not expensive at all.
- For Capturing image from real world we used Mobile CAM. Resolution is a fact but performance moderate so far.

We used several image descriptors for image detection and recognition. Nokia 5800 as webcam. We’ve developed circuit for switching purpose. Parallel port for interfacing system.

Check this video ……….

Click here Project AutoVEST” or,

# Window Function

**Posted:**May 3, 2009

**Filed under:**dsp, signal processing, window |

**Tags:**blackman, dsp, fswind, hamming, signal processing, window Leave a comment

Window function an important term in signal processing. It also called apodization or tapering function. It is useful for various operations, like separtion of band frequences etc.

Several different windows are available. Only two are worth using, the Blackman window and the Hamming window.

Blackman window are defined as,

**w(n)= a0 – a1*cos((2*pi*n)/(N-1)) + a2*cos((4*pi*n)/(N-1)); **

where, **0<=n<=N .**

and

**a0=(1-@)/2, a1=1/2, a2=@/2.** in blackman window **@** equal to 0.16

Now , in case of Hamming windows coefficient are computed from following equation,

**w(n)=0.54 – 0.46cos((2*pi*n)/N);** where, **0<=n<=N**.

Now for sample 74-point Blackman window’s time and frequency domain representation are given below,

For the same point Hamming window’s time and frequency domain is

In this case of blackman window Leakage factor is 0%, sidelobe attenuation -58.1dB and mainlobe width is (-3dB) : 0.0449.

for hamming, Leakage factor is 0.04%, sidelobe attenuation -42.5 dB and mainlobe width is (-3dB) : 0.0351.

After convolution this two windows we will get time and frequency domain like this,

Leakage factor is 0%, sidelobe attenuation -72.2dB and mainlobe width is (-3dB) : 0.0273.

so this result is more better than before. I called it FSwind ðŸ™‚