/SVM-Gaussian-Classification-FPGA

SVM Gaussian Classifier of 30x30 greyscale image on Verilog

Primary LanguageVerilog

The program computes 629 Ai's by multiplication of 900 <img src="https://latex.codecogs.com/gif.image?\large&space;\dpi{110}&space;e^{\frac{-\left|x_j-z_i_j\right|}{2\sigma^2}}" />



Here the z_ij is the (629 SVMs: Zi * 30x30 pixel set: Zij) and x_j (30x30 image vector: X)
sigma assumed to be 625
The weights are ai
and then add those Ai's to compute the final result i.e. classification of X
i[0,629]
j[0,899]


The following data are intialized in a file
	ai -- 629 Values stored in Logarithmic Number System (LNS)
		  weighted according to the 'weight' value specified ranging
		  from 0 to 30
	zi -- 629 SVMs each 900 pixel (8bit)
	xi -- Input image 900 pixel (8bit)
The same data is used by the Hardware Design Testbench (Verilog) to initilize 
 	ai and zi; and
	periodic entering of image X into the design