tiadalg_image_color_blending.h File Reference

This file defines the interface for image color blending to highlight pixel level detection derived by other algorithm such as semantic segmentation. More...

#include <stdint.h>

Go to the source code of this file.

Functions
int32_t	tiadalg_image_color_blending_cn (uint8_t *in_img[], int32_t in_img_width, int32_t in_img_pitch, int32_t in_img_height, int32_t num_planes, int32_t in_color_format, int32_t out_color_format, int32_t valid_region[4], uint8_t *det_plane, int32_t det_plane_pitch, float det_scale_fact, uint8_t *det_color_map_uv, uint8_t *out_img[], int32_t out_img_pitch)
	It does basic color blending to input image based on pixel level detection information provided by other module. More...
int32_t	tiadalg_image_color_blending_c66 (uint8_t *in_img[], int32_t in_img_width, int32_t in_img_pitch, int32_t in_img_height, int32_t num_planes, int32_t in_color_format, int32_t out_color_format, int32_t valid_region[4], uint8_t *det_plane, int32_t det_plane_pitch, float det_scale_fact, uint8_t *det_color_map_uv, uint8_t *out_img[], int32_t out_img_pitch)
	It does basic color blending to input image based on pixel level detection information provided by other module. More...

Detailed Description

This file defines the interface for image color blending to highlight pixel level detection derived by other algorithm such as semantic segmentation.

Version

0.1 Sep 2018 : Initial Code

Function Documentation

tiadalg_image_color_blending_cn()

int32_t tiadalg_image_color_blending_cn	(	uint8_t *	in_img[],
		int32_t	in_img_width,
		int32_t	in_img_pitch,
		int32_t	in_img_height,
		int32_t	num_planes,
		int32_t	in_color_format,
		int32_t	out_color_format,
		int32_t	valid_region[4],
		uint8_t *	det_plane,
		int32_t	det_plane_pitch,
		float	det_scale_fact,
		uint8_t *	det_color_map_uv,
		uint8_t *	out_img[],
		int32_t	out_img_pitch
	)

It does basic color blending to input image based on pixel level detection information provided by other module.

tiadalg_image_color_blending_cn

Parameters

in_img	[in] : list of input planes pointers
in_img_width	[in] : Input image width
in_img_pitch	[in] : Input image one line pitch
in_img_height	[in] : Input image height
num_planes	[in] : Num of planes in input image
in_color_format	[in] : Input color format. 0 means YUV 420 nv12 format, 1 means RGB de-ineterleaved
out_color_format	[in] : Output color format. 0 means YUV 420 nv12 format, 1 means RGB de-ineterleaved
valid_region	[in] : valid region to be blended. It is expectecin [TL_x, TL_y, BR_x, BR_y]. For non valid region input is copied to output
det_plane	[in] : Detection plane. Each pixel has values 0 to num_class-1
det_plane_pitch	[in] : Detection plane one line pitch
det_scale_fact	[in] : Detection plane values needs to be scaled by this factor
det_color_map_uv	[in] : Interleaved U and V for each detection class
out_img	[out] : Output image buffer
out_img_pitch	[in] : Output image one line pitch

Note

• Number of valid pointers in the list in_img should be equal to num_planes
• Dimension of det_plane should be same as first plane of input image
• Data range for /c det_plane is assumed to be [0 - num_class-1]
• valid element in /c det_color_map_uv, should be equal to num_class*2

tiadalg_image_color_blending_c66()

int32_t tiadalg_image_color_blending_c66	(	uint8_t *	in_img[],
		int32_t	in_img_width,
		int32_t	in_img_pitch,
		int32_t	in_img_height,
		int32_t	num_planes,
		int32_t	in_color_format,
		int32_t	out_color_format,
		int32_t	valid_region[4],
		uint8_t *	det_plane,
		int32_t	det_plane_pitch,
		float	det_scale_fact,
		uint8_t *	det_color_map_uv,
		uint8_t *	out_img[],
		int32_t	out_img_pitch
	)

It does basic color blending to input image based on pixel level detection information provided by other module.

tiadalg_image_color_blending_c66

Parameters

in_img	[in] : list of input planes pointers
in_img_width	[in] : Input image width
in_img_pitch	[in] : Input image one line pitch
in_img_height	[in] : Input image height
num_planes	[in] : Num of planes in input image
in_color_format	[in] : Input color format. 0 means YUV 420 nv12 format, 1 means RGB de-ineterleaved
out_color_format	[in] : Output color format. 0 means YUV 420 nv12 format, 1 means RGB de-ineterleaved
valid_region	[in] : valid region to be blended. It is expectecin [TL_x, TL_y, BR_x, BR_y]. For non valid region input is copied to output
det_plane	[in] : Detection plane. Each pixel has values 0 to num_class-1
det_plane_pitch	[in] : Detection plane one line pitch
det_scale_fact	[in] : Detection plane values needs to be scaled by this factor
det_color_map_uv	[in] : Interleaved U and V for each detection class
out_img	[out] : Output image buffer
out_img_pitch	[in] : Output image one line pitch

Note

• Currently only YUV420 nv12 format is supported for input and output for optimized kernel
• Number of valid pointers in the list in_img should be equal to num_planes
• Dimension of det_plane should be same as first plane of input image
• Data range for /c det_plane is assumed to be [0 - num_class-1]
• valid element in /c det_color_map_uv, should be equal to num_class*2