AI

CH-4 – Wavelets Feature Extraction – Part-7

Posted on Author Ali Imran Comment(0)

Important Note: This article is part of the series in which TechReport.us discuss theory of Video Stream Matching.

There are two Algos are discus. One for wavelets coefficient values and second one is according to paper.[6]

4.4.4.1 Algorithm 8

Input:

BMP Image

Output:

Array Of Numerical values define the Wavelets Coefficient values property.

Working:

It takes an Image and converts it into gray scale if it is not already in gray scale. For wavelets it apply wavelets HARR on image and return coefficient values.

Step 0 : Start

Step 1 : Takes input an Image (im).

Step 2 : If (im) in gray scale

If not

then

convert (im) into gray scale.

Step 3 : (im1) = Apply Padding to make it into power of two (im)

Step 4 : (I , J) = Size (im1)

Step 5 : Apply Filter(im1,’gaussian’,[5 5],1.4)

Step 6 : [ cA , cH , cV , cD ] = dwt2(im1,’haar’); // Apply HAAR transformation

Step 7 : t = graythresh(cA); // Apply Threshold on cA

Step 8 : Loop1: for I =rows of (cA)

Step 9 : Loop2: for J = columns of (cA)

Step 10 :

if cA(i,j) < t

cA(i,j) =0;

end

Step 11 : end Loop1

Step 12 : end Loop2

Step 13 : Return (cA)

Step 14 : Stop

4.4.4.2 Algorithm 9

Input:

BMP Image

Output:

Array Of Numerical values define the Wavelets positive and negative values property.

Working:

It takes an Image and converts it into gray scale if it is not already in gray scale. For wavelets it apply wavelets HARR on image and return matrix which is consist on each tiles positive values and negative values.

Step 0 : Start

Step 1 : Takes input an Image (im).

Step 2 : If (im) in gray scale

If not

then

convert (im) into gray scale.

Step 3 : (im1) = Apply Padding to make it into power of two (im)

Step 4 : (I , J) = Size (im1)

Step 5 : Apply Filter(im1,’gaussian’,[5 5],1.4)

Step 6 : [ cA , cH , cV , cD ] = dwt2(im1,’haar’); // Apply HAAR transformation

Step 7 : imadd = cH+cV; // combine horizontal and vertical effects

Step 8 : [r c] = size(imadd);

nc=0;

pc=0;

wp=1;

wn=1;

wh=1;

Step 9 : Loop1: for I =1:16:r // make tiles of image (imadd)

Step 10 : Loop2: for J = 1:16:c

Step 11 :

if imadd(g,k)>0

pc=pc+1;

else

nc=nc+1;

end

wpmat(wh,wp)=pc; // after each tile do this

wnmat(wh,wn)=nc;

wp=wp+1;

wn=wn+1;

pc=0;

nc=0;

Step 12 : end Loop1

wh=wh+1;

wp=1;

wn=1;

Step 13 : end Loop2

Step 14 : out = [ wpmat(:); wnmat(:)];

Step 15 : Return (out)

Step 16 : Stop

Ali Imran
Over the past 20+ years, I have been working as a software engineer, architect, and programmer, creating, designing, and programming various applications. My main focus has always been to achieve business goals and transform business ideas into digital reality. I have successfully solved numerous business problems and increased productivity for small businesses as well as enterprise corporations through the solutions that I created. My strong technical background and ability to work effectively in team environments make me a valuable asset to any organization.
https://ITsAli.com

Related Articles
AI

CH-4 – Edge Feature Extraction – Part-5

Posted on Author Ali Imran

Algorithm 3 extracts edge map features from BMP images. It begins by converting the image to grayscale if necessary and applies the Canny algorithm, returning the image edges. The process involves steps like Gaussian filtering to remove noise, Sobel operator application in horizontal and vertical directions, and angle calculation for each pixel. The resulting edges are thresholded and subjected to double thresholding for refinement. Finally, the processed image with edge maps is returned.

Algorithm 4 operates similarly but focuses on extracting edge angles instead of edge maps. It follows a comparable process, including Gaussian filtering, Sobel operator application, and angle calculation, but stores the occurrences of each angle in tiles of the image, returning an array defining the edge angles.
AI

CH-4 – Features Extraction – Part-4

Posted on Author Ali Imran

Algorithm 2 extracts histogram features from BMP images. It begins by taking an image as input and converts it to grayscale if necessary. The image is divided into tiles, and histograms are calculated for each tile. These histograms are stored in a 2D array, representing the result. The algorithm iterates through tiles, extracting sub-images, calculating histograms, and storing them in the result array. Finally, the resulting array of numerical values defining the histogram is returned, providing essential feature extraction for subsequent analysis in video stream matching.
AI

CH-3 – Image Analysis – Part-5

Posted on Author Ali Imran

Wavelet feature extraction involves transforming images into wavelet coefficients. The number of data elements per image is determined based on the image’s rows and columns, typically brought to a power of two. The total number of values after applying the wavelet is calculated. Classification employs a Dual Statistical Approach, combining parametric and non-parametric methods like MEAN, NORM, Variance, and KS TEST. The Minimum Distance Classifier, utilizing Euclidean, Normalized Euclidean, and Mahalanobis distances, aids in classifying unknown image data by minimizing distances between data and classes in multi-feature space.

Leave a Reply