Bilateral Filter - 2023.2 English

In general, any smoothing filter smoothens the image which will affect the edges of the image. To preserve the edges while smoothing, a bilateral filter can be used. In an analogous way as the Gaussian filter, the bilateral filter also considers the neighboring pixels with weights assigned to each of them. These weights have two components, the first of which is the same weighing used by the Gaussian filter. The second component takes into account the difference in the intensity between the neighboring pixels and the evaluated one.

The bilateral filter applied on an image is:

Where
and  is a gaussian filter with variance .

The gaussian filter is given by:

API Syntax

template<int FILTER_SIZE, int BORDER_TYPE, int TYPE, int ROWS, int COLS, int NPC=1, int XFCVDEPTH_IN_1 = _XFCVDEPTH_DEFAULT, int XFCVDEPTH_OUT_1 = _XFCVDEPTH_DEFAULT>
void bilateralFilter (
xf::cv::Mat<TYPE, ROWS, COLS, NPC, XFCVDEPTH_IN_1>& src,
xf::cv::Mat<TYPE, ROWS, COLS, NPC, XFCVDEPTH_OUT_1>& dst,
float sigma_space, float sigma_color )

Parameter Descriptions

The following table describes the template and the function parameters.

Table 300 Table bilateralFilter Parameter Description
Parameter	Description
WINDOW_SIZE	Filter size. Filter size of 3 (XF_FILTER_3X3), 5 (XF_FILTER_5X5) and 7 (XF_FILTER_7X7) are supported
BORDER_TYPE	Border type supported is XF_BORDER_CONSTANT
TYPE	Input and output pixel type. Only 8-bit, unsigned, 1 channel, and 3 channels are supported (XF_8UC1 and XF_8UC3)
ROWS	Maximum height of input and output image.
COLS	Maximum width of input and output image (must be multiple of 8, for 8-pixel operation)
NPC	Number of pixels to be processed per cycle; this function supports XF_NPPC1 and XF_NPPC4.
XFCVDEPTH_IN_1	Depth of input image
XFCVDEPTH_OUT_1	Depth of output image
src	Input image
dst	Output image
sigma_space	Standard deviation of filter in spatial domain
sigma_color	Standard deviation of filter used in color space

Resource Utilization

The following table summarizes the resource utilization of the kernel in different configurations, generated using Vivado HLS 2019.1 version tool for the Xczu9eg-ffvb1156-1-i-es1 FPGA, to progress a grayscale HD (1080x1920) image.

Table 301 Table bilateralFilter Resource Utilization Summary
Operating Mode	Filter Size	Operating Frequency (MHz)	Utilization Estimate
Operating Mode	Filter Size	Operating Frequency (MHz)	BRAM_18K	DSP_48Es	FF	LUT
1 Pixel	3x3	300	6	22	4934	4293
	5x5	300	12	30	5481	4943
	7x7	300	37	48	7084	6195

Table 302 Table bilateralFilter Resource Utilization Summary
Operating Mode	Filter Size	Operating Frequency (MHz)	Utilization Estimate
Operating Mode	Filter Size	Operating Frequency (MHz)	BRAM_18K	DSP_48Es	FF	LUT
1 Pixel	3x3	300	12	32	8342	7442
	5x5	300	27	57	10663	8857
	7x7	300	49	107	12870	12181

Performance Estimate

The following table summarizes a performance estimate of the kernel in different configurations, as generated using Vivado HLS 2019.1 tool for Xczu9eg-ffvb1156-1-i-es1 FPGA, to process a grayscale HD (1080x1920) image.

Table 303 Table 35. bilateralFilter Function Performance Estimate Summary
Operating Mode	Filter Size	Latency Estimate
		300 MHz
		Max Latency (ms)
1 pixel	3x3	7.18
	5x5	7.20
	7x7	7.22

Deviation from OpenCV

Unlike OpenCV, Vitis Vision only supports filter sizes of 3, 5 and 7.