# US 6,731,794 B2 · Method and apparatus for estimating true color values for saturated color values in digitally captured image data

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

- Number: US 6,731,794 B2
- Document ID: US6731794B2
- Status: Granted
- Kind: B2
- Application number: 09/834,807
- Filed: 2001-04-12
- Published: 2002-11-28
- Granted: 2004-05-04
- Earliest claimed: 2001-04-12
- Assignees: Hewlett-Packard Development Company, L.P.
- Inventors: Xuemei Zhang; David H Brainard
- Web page: https://patentgrep.com/patent/US6731794B2

## CPC

- G06T 7/90: Determination of colour characteristics
- H04N 2101/00: Still video cameras
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- H04N: PICTORIAL COMMUNICATION, e.g. TELEVISION

## Abstract

A system and method of adjusting a saturated color value in a digital color image by estimating a true color value for the saturated color value using a statistical technique where a prior statistical distribution of color values representing the probability of captured responses of an image capture device is used to determine a posterior distribution of possible true color values for the saturated color values. The method is preformed by obtaining a first statistical distribution of color values associated with all pixel locations in the digital image that represents captured responses of an image capture device. A saturated color value of a first color component and its location is detected in the digital data. Based on the other color component color values at the detected location, a second statistical distribution is derived from the first distribution that represents a distribution of possible true color values for the detected saturated color value. The mean of the second distribution provides an estimation of the true color value for adjusting the detected saturated color value.

## Claims

### Claim 1

A method of estimating a true color value for a saturated color value in digital color image data, the method comprising:
obtaining a first distribution associated with all pixel locations in the image data that represents a probability of captured responses of an image capture device;
detecting a saturated color value of a first color component and its location and the image data and for each detected value:
deriving a second distribution representing possible true color values of the detected value from the first distribution based on at least one other color component color value at the detected location; and
determining the mean of the second distribution to obtain the estimation of the true color value.

### Claim 2 (depends on 1)

The method as described in claim 1 wherein obtaining the first distribution comprises assuming that the first distribution is a multivariate normal distribution.

### Claim 3 (depends on 2)

The method as described in claim 2 further comprising parameterizing the multivariate normal distribution by a mean vector and covariance matrix of non-saturated color values of all pixel locations in the image data.

### Claim 4 (depends on 1)

The method as described in claim 1 wherein obtaining the first distribution comprises assuming that the first distribution is a non-normal distribution.

### Claim 5 (depends on 1)

The method as described in claim 1 wherein obtaining the first distribution comprises theoretically determining the first distribution by obtaining a distribution of color values from a collection of object surface reflectance measurements, scene illuminants, and camera sensor spectral sensitivities.

### Claim 6 (depends on 1)

The method as described in claim 1 wherein obtaining the first distribution comprises estimating the first distribution by analyzing a large collection of predetermined images to generate an estimated first distribution of the predetermined images.

### Claim 7 (depends on 1)

The method as described in claim 1 wherein obtaining the first distribution comprises estimating the first distribution from the digital color image data.

### Claim 8 (depends on 1)

The method as described in claim 1 wherein the image data is demosaiced digital data.

### Claim 9 (depends on 1)

The method as described in claim 1 wherein the image capture device corresponds to the actual device used to obtain the image data.

### Claim 10 (depends on 1)

The method as described in claim 1 wherein the step of deriving a second distribution further comprises deriving a second distribution from the first distribution based on at least one other color component color value of a different type at the detected location.

### Claim 11

A system of estimating a true color value for a saturated color value in digital color image data comprising:
a means for determining a first distribution associated with all pixel locations in the image data that represents a probability of captured responses of an image capture device;
a means for detecting a saturated color value in the image data and its location;
a means for deriving a second distribution representing possible true color values of the detected value from the first distribution based on at least one color value at the saturated pixel locations;
a means for processing the second distribution to obtain the estimated true color value by determining a mean of the second distribution.

### Claim 12 (depends on 11)

The system as described in claim 11 wherein the captured responses used to determine the first distribution are spectral sensitivities, scene illuminants, and a representative collection of object surface reflectance measurements.

### Claim 13 (depends on 11)

The system as described in claim 11 wherein the captured responses used to determine the first distribution are the non-saturated pixel location color values on the image data.

### Claim 14 (depends on 11)

The system as described in claim 11 wherein the captured responses used to determine the first distribution are color values of digital image data from a collection of representative images.

### Claim 15 (depends on 11)

The system as described in claim 11 wherein the first distribution is assumed to be a multivariate normal distribution which is characterized by a mean vector and a covariance matrix which are estimated using the non-saturated pixel location color values of the digital image.

### Claim 16 (depends on 11)

The system as described in claim 11 wherein the at least one color value is non-saturated.

### Claim 17 (depends on 11)

The system as described in claim 11 wherein the at least one color value is saturated.

### Claim 18 (depends on 11)

The system as described in claim 11 wherein the first distribution is assumed to be a non-normal distribution.

### Claim 19 (depends on 11)

The system as described in claim 11 wherein the determination means, the detection means, the derivation means and the processing means are implemented by software, hardware, or any combination of software and hardware.

### Claim 20 (depends on 11)

The system as described in claim 11 wherein the determination means, the detection means, the derivation means and the processing means are implemented within a computing system having at least a central processing unit (CPU) and a memory and storage area for storing digital data.

### Claim 21 (depends on 11)

The system as described in claim 11 wherein the at least one color value comprises a different type of color value.
