# US 2019/0057491 A1 · Skin map-aided skin smoothing of images using a bilateral filter

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2019/0057491 A1
- Document ID: US20190057491A1
- Status: Published application
- Kind: A1
- Application number: 16/079,308
- Filed: 2016-03-24
- Published: 2019-02-21
- Earliest claimed: 2016-03-24
- Assignees: Intel Corporation
- Applicants: Intel Corporation
- Inventors: Liu YANG; Weike CHEN; Lin XU
- Web page: https://patentgrep.com/patent/US20190057491A1

## CPC

- G06T 5/70: Denoising; Smoothing
- G06T 5/20: using local operators
- G06T 5/77: Retouching; Inpainting; Scratch removal
- G06T 2207/20028: Bilateral filtering
- G06T 2207/30088: Skin; Dermal
- G06T 2207/30201: Face
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

Skin smoothing is applied to images using a bilateral filter and aided by a skin map. In one example a method includes receiving an image having pixels at an original resolution. The image is buffered. The image is downscaled from the original resolution to a lower resolution. A bilateral filter is applied to pixels of the downscaled image. The filtered pixels of the downscaled image are blended with pixels of the image having the original resolution, and the blended image is produced.

## Claims

### Claim 1

A method comprising:
receiving an image having pixels at an original resolution;
buffering the image;
downscaling the image from the original resolution to a lower resolution;
applying a bilateral filter to pixels of the downscaled image;
blending the filtered pixels of the downscaled image with pixels of the image having the original resolution; and
producing the blended image.

### Claim 2 (depends on 1)

The method of claim 1, further comprising generating a skin map of the image having the original resolution and wherein blending comprises using pixels of the filtered image for areas indicated as skin by the skin map.

### Claim 3 (depends on 2)

The method of claim 2, wherein generating a skin map comprises generating a weight for each pixel indicating the probability that each pixel corresponds to skin and wherein using pixels of the filtered image comprises applying the weight to combine a pixel of the image at the original resolution with a pixel of the filtered image.

### Claim 4 (depends on 3)

The method of claim 3, wherein applying the weight comprises multiplying a pixel of the filtered image by a first factor based on the weight, multiplying a corresponding pixel of the image of the original resolution by a second factor based on the weight, and then adding the two factored pixels.

### Claim 5 (depends on 2)

The method of claim 2, wherein blending comprises using only pixels of the image having the original resolution for areas indicated as not skin by the skin map.

### Claim 6 (depends on 1)

The method of claim 1, wherein downscaling comprises downscaling the image to one half the width and one half the height of the original resolution.

### Claim 7 (depends on 1)

The method of claim 1, further comprising generating a skin map of the image having the original resolution and wherein applying a bilateral filter comprises applying a bilateral filter only to areas indicated as skin by the skin map.

### Claim 8 (depends on 1)

The method of claim 1, wherein applying a bilateral filter comprises applying an edge preserving bilateral filter with a local histogram optimization.

### Claim 9

An apparatus comprising:
a downscaler to receive an image having pixels at an original resolution and to downscale the image from the original resolution to a lower resolution;
a buffer to store the downscaled image;
a bilateral filter to filter pixels of the lower resolution image;
a blender to blend the pixels of the filtered image having the lower resolution with pixels of the image having the original resolution; and
a buffer to store the blended image.

### Claim 10 (depends on 9)

The apparatus of claim 9, further comprising a skin map generator to generate a skin map of the image having the original resolution and wherein the blender uses pixels of the filtered image for areas indicated as skin by the skin map.

### Claim 11 (depends on 9)

The apparatus of claim 9, further comprising skin map generator to generate a skin map of the downscaled image having a weight for each pixel indicating the probability that each pixel corresponds to skin and wherein the blender applies the weights to combine a pixel of the image at the original resolution with a pixel of the filtered image.

### Claim 12 (depends on 11)

The apparatus of claim 11, wherein applying the weight comprises multiplying a pixel of the filtered image by a factor based on the weight, and then adding the factored pixel to a corresponding pixel of the original resolution image.

### Claim 13 (depends on 9)

The apparatus of claim 9, wherein the bilateral filter comprises an edge preserving bilateral filter with a local histogram optimization.

### Claim 14

A video terminal comprising:
a camera to generate a sequence of images having pixels at an original resolution;
a downscaler to receive the images having pixels at the original resolution and to downscale the images from the original resolution to a lower resolution;
a buffer to store the downscaled image;
a bilateral filter to filter pixels of the downscaled image;
a blender to blend the filtered pixels of the downscaled image with pixels of the image having the original resolution; a buffer to store the blended images; and
a display to present the sequence of images after blending.

### Claim 15 (depends on 14)

The terminal of claim 14, further comprising a skin map generator to generate a skin map of the downscaled image and wherein the blender uses pixels of the filtered image for areas indicated as skin by the skin map.

### Claim 16 (depends on 14)

The terminal of claim 14, further comprising a skin map generator to generate a skin map of the downscaled image having a weight for each pixel indicating the probability that each pixel corresponds to skin and wherein the blender applies the weights to combine a pixel of the image at the original resolution with a pixel of the filtered image.

### Claim 17 (depends on 16)

The terminal of claim 16, wherein the blender applies a low threshold to the weights so that if a weight is below the low threshold for a pixel, then the pixel at the original resolution is not combined with the corresponding pixel of the filtered image.

### Claim 18 (depends on 16)

The terminal of claim 16, wherein the blender applies a high threshold to the weights so that if a weight is above the high threshold for a pixel, then the pixel a the original resolution is replaced with the corresponding pixel of the filtered image.

### Claim 19 (depends on 14)

The terminal of claim 14, wherein applying the weight comprises multiplying a pixel of the filtered image by a factor based on the weight, and then adding the factored pixel to a corresponding pixel of the original resolution image.

### Claim 20 (depends on 14)

The terminal of claim 14, further comprising a communications chip to transfer the sequence of images from the terminal.
