#Bilateral Filter

Bilinear filter is a nonlinear, edge-preserving smoothing filter that is often used to smooth pictures. It takes the weighted average of the intensity values of nearby pixels in each input image, and it calculates the weights by the Euclidean distance between pixels and color intensity differences, and finally obtains the picture processing effect of preserving the edges and smoothing the similar regions.

#Operator Effect

Input Image	Parameter	Output Image
	kernelSize = 5 sigmaColor = 15.0f sigmaSpace = 2.0f borderType = HB_VP_BORDER_REPLICATE

#Principle

The main computational process of bilinear filtering is to convolve the input picture using a filter kernel, the main equation is as follows:

$dst(x)=\frac{1}{\sum_{x_i\in\Omega}{f_r(x,xi)g_s(x,xi)}}\displaystyle\sum_{x_i\in\Omega}{src(x_i)f_r(x,xi)g_s(x,xi)}$

Among them, $dst$ is the output image, $src$ is the input image, $x$ is the currently computed pixel, and $\Omega$ is the current filter window.

$f_r(x,xi)=e^{-\frac{1}{2}(\frac{\lvert src(xi)-src(x)\rvert}{\sigma_r})^2}$

$g_s(x,xi)=e^{-\frac{1}{2}(\frac{\lVert xi-x \rVert}{\sigma_d})^2}$

Among them, $\vert$ is the absolute value, $\Vert$ is the Euclidean distance, $\sigma_r$ is the color sigma parameter for smoothing intensity differences across the filter window, and $\sigma_d$ is the spatial sigma parameter for smoothing differences in spatial coordinates across the window.

#API Interface

For detailed interface information, please refer to hbVPBilateralFilter.

#Usage