Important Note: This article is part of the series in which TechReport.us discuss theory of Video Stream Matching.
3.2.2.3 – Parametric Data and Non Parametric Data
There is data which shows the same distribution and has distribution called parametric data. In this type of data, distance form mean on either side remain same. But in non-parametric data distance from mean is not same. It is data which has no specific distribution. Simply this type is called distribution free data. [25, 26, 27]
This mean frame histogram data is non parametric data so that is why using in it KS Test.
3.3 – Features Extraction
There are four types of features selected for this system.[6]
- Histogram
- Edge
- Slope
- Wavelets
Now these features will discuss one by one below.
3.3.1 – Histogram Feature Extraction
Method of Histogram feature extraction is same as above for Mean Frame. Same formulas are used in it. It provide additional information after calculating histogram tile by tile number of tiles and also number of bins per window smaller size of bins more elements in feature Sub-Vector.[6]
After calculating the result, data arrange in column form means number of Tiles and number of columns are same.
W1 W2 W3 W4
1 2 3 4
2 3 1 2
3 4 2 1
4 1 4 3
So above data shows how data arrange from W1 to W4. Also shows the number of Tiles are 4 and below it shows that Bins are 4 and each window has four elements.
Here one problem lies. Even values in each window are same and they have different distribution. But if calculate there mean then outcome is same values.
Mean of W1 W2 W3 W4
Mean = 2.5 2.5 2.5 2.5
So to solve this problem, values of each window is multiplying with a column vector which contains the values from 1 to number of bins. For example if Number of Bins are 4 then column vector consist on.
| 1 |
| 2 |
| 3 |
| 4 |
Now with each window matrix multiplying with this matrix this process is called weighting the matrix.
CM W1 ResW1
1 1 1
2 X 2 = 4
3 3 9
4 4 16
CM w2 ResW2
1 2 2
2 X 3 = 6
3 4 12
4 1 4
Mean of ResW1 ResW2
Mean= 7.5 6
Now this process make able to system to get different results. It is very unique that system got same results. If this may happen in data any where then it has solution that instead of using mean use variance of data.
According to experiments the pair of mean values and variance values is not same.
For example
W1 W2 W3 W4
2 4 4 2
4 2 6 6
6 6 2 4
Now multiplying W1, W2, W3, W4 with Mat
Mat
| 1 |
| 2 |
| 3 |
Then resultant matrixes are.
RW1 RW2 RW3 RW4
2 4 4 2
8 4 12 12
8 18 6 12
Now mean and variance of above matrixes are,
RW1 RW2 RW3 RW4
Mean = 9.33 8.66 7.33 8.66
Variance = 65.33 65.33 17.33 33.33
Here getting information that pair of Mean and Variance is not same. So this result is further used for better decision making.
