1. Field of the Invention
The invention pertains to the field of digital imaging and, more particularly, to white balance control for adjustment of color or shade under variable lighting conditions.
2. Discussion of the Related Art
Many digital cameras and imaging devices use an array of charge-coupled device (CCD) detectors to capture the image. Each CCD detector accumulates a charge when exposed to light. The magnitude of charge represents the intensity of the light to which the CCD detector has been exposed. Each CCD detector is coupled to an analog to digital converter (A/D) that produces a digital signal representing the charge and intensity of light received by the CCD detector. Color images are captured by placing a color filter before each CCD detector. For example, a grouping of CCD color detectors may include two green detectors, a red detector and a blue detector. The color composition of an image is captured by taking the digital value of each detector according to color type. By way of example, U.S. Pat. No. 5,995,142 issued to Matsufune shows circuitry for capturing color images where analog signals from the CCD detectors are sorted into color channels, subjected to analog processing via a variable gain amplifier at the channel level, and converted to digital signals.
Complimentary metal oxide semiconductor (CMOS) detectors may be used in place of CCD detectors. CMOS detectors differ from CCD detectors in that CMOS detectors provide a continuous signal representing light intensity, whereas CCD detectors are integrating devices. Collectively, CCD detectors, CMOS detectors, and other useful forms of photodetectors or photodiodes are referred to herein as “detectors.
It is problematic in the art of digital photography that the digital cameras are used under variable light conditions. The signal received at each CCD detector may vary, for example, depending upon the type of light source even when the total intensity of light between two different sources is the same. These variations occur because different light sources differ in their emission spectra. As compared to natural sunlight, the emissions from a tungsten source contain more red light than blue or green light. If a digital camera is tuned or calibrated for natural sunlight, the resulting image under a tungsten source has an undesirable reddish/orange appearance. In addition, even ambient light from the same type of source may vary in intensity and composition, for example, as natural sunlight that is filtered by clouds or by atmosphere according to the season of the year. These problems are mitigated by providing digital cameras with white balance adjustments.
White balance is achieved when the image of a neutral white object produces equalized or normalized signals in each of the red, blue and green CCD detector channels. A digital camera that is tuned or calibrated to achieve white balance under one set of illumination conditions is not necessarily calibrated for a different set of illumination conditions. As shown in the '142 patent to Matsufune, one method for overcoming this problem is to alter the gain of the variable gain amplifiers according to principles of calibration involving the adjustment of variable gain coefficients. The '142 patent shows a system for scanning an image according to a calibration algorithm that follows a blackbody radiation pattern. The blackbody radiation pattern is used to adjust or select the scan area that is used for white balance adjustment purposes, in combination with stored standardized calibration data representing the color white.
Yet another white balance adjustment technique is described in U.S. Pat. No. 6,411,331 issued to Sansom-Wai et al. Circuitry is coupled to a CCD array for generating uncorrected digital image data including a chromaticity value and a luminance value for each of a plurality of pixels. A control circuit maps a two dimensional representation of the uncorrected image to select a spatial region that is used for calibration purposes in making the white balance adjustment.
U.S. Pat. No. 6,249,323 issued to Van Der Voort describes non-scanning circuitry that sets variable gain coefficients by equalizing chromiluminsence signals to achieve white balance. The variable gain coefficients may be manually set through adjustment of a variable resistor, or they may be set automatically in response to input signals, e.g., from an imaged calibration target of the color white or a specified shade of gray. Gain correction is performed according to a mathematical algorithm that exponentially and proportionally relate the input signals to stored reference signals for the calibration target.
U.S. Pat. No. 6,038,339 issued to Hubel et al. describes a variety of mathematical techniques that may be used to calculate white balance and describes improvements that pertain to correcting color of the image based on the illumination type. A correlation matrix memory or associative matrix memory is used to achieve identical results to the known color-in-perspective method. The methods are improved by adding Bayesian or other correlation statistics. The correlation matrix memory is built to correlate the data from any picture image to reference images under a range of illumination type. When a camera, scanner, or the like, produces a picture image, the data is converted to chromaticity and a vector is created corresponding to the values existing in the scene. This vector is multiplied by each column in the correlation matrix to establish a new matrix. Each column is then summed, and the resulting values form a vector that is compared to stored reference information representing the likelihood of each reference source being the illumination type for the scene. The vector values can be density plotted where each value is plotted at the chromaticity of the illumination for that particular column. From this plot normal statistical methods are used to estimate the likely illumination type of the scene. Greatly reduced computational requirements result because simple binary matrix algorithms replace complex geometrical calculations.
None of the foregoing references address a more practical problem that is commonly encountered by photographers. The quality of photographic images is enhanced by calibrating digital cameras under actual conditions of illumination encountered as the photographs are being taken. Calibration of this type requires that the photographer have access to a calibration target. The calibration target is frequently specified as white paper having 18% gray. The target is subject to degradation under conditions of actual use, such as contamination with grime, creasing, or tearing. Nothing in the art addresses the practical problem of having to produce a pristine calibration target in the field.
Features described herein advance the art and overcome the problems outlined above by providing an imaging device, such as a digital still camera or digital motion picture camera, with white balance adjustment logic that permits the use of alternative calibration targets other than the standard calibration target that is specified by the manufacturer for use in adjusting white balance.
According to one aspect, the imaging device includes image capture circuitry that is configured to produce captured image signals. The captured image signals are, respectively, obtained as a primary image of a standard target that comports with manufacturer specifications for use in making white balance adjustments, and a secondary image of a non-standard target. The primary image and the secondary image are obtained under identical illumination conditions. The non-standard target is, for example, the palm of a photographer's hand, or a camera lens cover. Signal processing circuitry processes the primary image to obtain a first reference metric, such as a variable gain coefficient for use on at least one of a red, blue or green channel (RGB). It will be appreciated that additional channels may be utilized, for example, by adding a yellow or orange channel to RGB. The signal processing circuitry processes the secondary image, e.g., in an identical manner with respect to the primary image, to obtain a second reference metric. The photographer is then able to obtain a field image of the non-standard target under different illumination conditions. The signal processing circuitry processes the field image to obtain a third reference metric and relates the third reference metric to at least one of the first reference metric or the second reference metric, to calibrate the camera in adjusting white balance.
By way of example, the reference metrics may all be variable gain coefficients for use in adjusting the output of a charge coupled device array. A variable gain coefficient is hereby defined to include a value, or a digital or analog representation of a value, that may be input to a variable gain amplifier for use in adjusting the output of the variable gain amplifier.
In one embodiment, the signal processing circuitry may be operable for determining at least one variable gain coefficient for the field image, determining at least one variable gain coefficient for the primary image, relating the variable gain coefficient for the field image to the variable gain coefficient for the primary image to produce an adjusted variable gain coefficient, and supplying the adjusted variable gain coefficient to the circuitry for use in adjusting white balance.
Another embodiment pertains to a method for adjusting white balance in an imaging device. The method includes a step of capturing image signals from a field image of a non-standard calibration target, and a primary image of a standard calibration target, to produce a captured field image and a captured primary image. The captured field image is related to the captured primary image to provide calibration for adjustment of white balance. White balance is adjusted by applying the calibration, for example, by enabling an adjusted variable gain coefficient on a variable gain amplifier.
In one aspect, the step of enabling an adjusted variable gain coefficient may include calculating a relationship between a variable gain coefficient for the field image and a variable gain coefficient for the primary image. This may be accomplished by determining at least one variable gain coefficient for the field image, determining at least one variable gain coefficient for the primary image, relating the variable gain coefficient for the field image to the variable gain coefficient for the primary image to produce the adjusted variable gain coefficient, and supplying the adjusted variable gain coefficient to the signal processing circuitry.
Upon reading the following detailed description, in addition to the accompanying drawings, those skilled in the art will appreciate that additional features and advantages are disclosed.
According to one embodiment, a user chooses a white balance technique from a list of white balance menu options 118. The signal processing circuits 116 are able to select from a variety of variable gain coefficient calculation techniques or functions according to the white balance menu options. For example, white balance menu options 118 present the user with selections to identify (1) the white balance control technique, and (2) the type of light source for a particular white balance control technique. The variable gain coefficients are generated by the signal processing circuits 116 according to the user-selected option from white balance menu options 118 and enabled on the variable gain amplifiers 110, 112, and 114. By way of example, the signal processing circuits 116 are provided with predetermined variable gain coefficients that achieve white balance under presumed average conditions of illumination for menu-selectable options including sunlight, tungsten lighting, and/or fluorescent lighting.
In one embodiment, the white balance menu options 118 permit the user to achieve white balance using a primary standard target 102 under ambient light conditions. The user manipulates camera 100 to obtain a primary image 122 from the primary standard target 102, which comports with manufacturer-specified standardization requirements for use as a white balance calibration tool, e.g., a white target with 18% gray. The signal processing circuits 116 access a white balance processing algorithm 124 to calculate appropriate variable gain coefficients, PC, that are enabled on the variable gain amplifiers 110, 112, and 114 to achieve white balance. Data from the signal processing circuits 116, and/or calculation results from white balance processing algorithm 124 including the variable gain coefficients, are stored as a stored standard target image 126. The white balance processing algorithm 124 is implemented using digital circuitry and/or analog circuitry, and/or using a processor configured with program instructions.
In yet another embodiment, after obtaining the primary image 122, the user manipulates camera 100 to obtain a secondary image 128 from the secondary non-standard target 104. The primary image 122 and the secondary image 128 are captured in close time proximity to one another and with similar positioning of camera 100 to assure that the respective images 122, 128, are exposed to the same intensity of light under the same type of light source. The secondary non-standard target 104 need not comport with manufacturer-specified standardization requirements for use as a white balance calibration target, and may be any object other than the primary standard target 102. By way of example, the secondary non-standard target 104 is the palm of the user's hand, a camera lens cover, a wallet, or a painted wall in a photography studio. The signal processing circuits 116 access the white balance processing algorithm 124 to generate appropriate variable gain coefficients that are enabled on the variable gain amplifiers 110, 112, and 114. Data from the signal processing circuits 116, and/or calculation results of white balance processing algorithm 124 including the variable gain coefficients, are stored as a stored secondary non-standard target image 132. In this embodiment, the white balance processing algorithm 124 determines the channel-specific variable gain coefficients SC that achieve white balance for the secondary non-standard target 104 and correlates these coefficients to the variable gain coefficients PC for the primary standard target, e.g., by determining a channel-specific signal multiplier MC as:
MC=ƒ1(PC,SC), (1)
where PC is the channel-specific variable gain coefficient for the primary standard target 102 under a condition of illumination, SC is the channel-specific variable gain coefficient for the secondary standard target image under the same condition of illumination, and ƒ1 is a function of PC and SC relating the output of detector array 106 to different intensities of illumination. For example, ƒ1 is a straight ratio or proportion of PC/SC, where the detector array 106 provides a charge output that is linearly related to the intensity of light contacting the detector array 106. The function ƒ1 can be an empirical correlation relating MC as a ratio, product, sum or difference of PC and SC. For example, ƒ1 can be a least square fit of empirical data for an illumination condition under tungsten light. A different function ƒ1 can be used for illumination conditions under direct sunlight, shade, or fluorescent light. The function ƒ1 can differ for each or the R, G and B channels. Signal processing circuits 116 enable the variable gain coefficients PC on the corresponding variable gain amplifiers 110, 112 and 114.
The foregoing calibration technique merely determines a relationship between PC and SC under uniform conditions of illumination. For example, the uniform conditions of illumination may occur by obtaining the primary image 122 and the secondary image 128 in close time proximity of less than from one to five minutes under lighting from a fluorescent source 134. Alternatively, fluorescent source 134 may be replaced by sunlight or by tungsten lighting, for example. Once the value SC has been obtained, there is no longer any need for the primary standard target 102.
In this embodiment, the user travels to a different location, for example, one that is illuminated by a different type of light other than fluorescent source 134. At this new location, the user accesses the white balance menu options 118 to select a field calibration mode. In this mode, the user further chooses from illumination source options, such as direct daylight, tungsten, fluorescence, and shade, which all are associated with different calibration functions described below. The user obtains field image 135 of secondary non-standard target 104, which is submitted to signal processing circuits 116 and white balance processing algorithm 124 to obtain channel-specific variable gain coefficients FC that achieve white balance on the basis of secondary non-standard target 104 in the field. The field image 135 and/or variable gain coefficients FC are stored as field correction data 136.
In the embodiment that uses the above Equation (1), signal processing circuitry 116 adjusts the variable gain coefficients PC applied to the variable gain amplifiers the 110, 112, 114 by enabling PA in place of PC according to Equation (2):
PA=MCFC, (2)
where PA is a channel-specific variable gain coefficient calibrated for field conditions of ambient light, as measured in field image 135.
In another embodiment, a function ƒ2 is empirically derived as a correlation to adjust white balance between different intensities for each channel, e.g., by using a factory-calibrated least squares algorithm relating PA as a function of FC and SC:
PA=ƒ2(FC,SC), (3)
where ƒ2 is an empirical function relating FC and SC to PA. It will be appreciated that function f2 can be an empirical function relating PA as a ratio, product, sum or difference of FC and SC. There may be as many types of function f2 as there are for function f1, e.g., channel-specific functions for conditions of direct sunlight, shade, tungsten light and fluorescent light.
Here it is also helpful to observe, in an ideal sense:
PC=ƒ2(FC,SC) (when FC=SC), (4)
which relationship may be used to confirm proper calibration of the camera 100 at the time of storing the stored secondary standard target image 132. If PC determined by Equation (4) differs from observed PC by more than a small amount, e.g., five percent, camera 100 may issue a user warning that calibration has failed.
With field calibration complete, the user accesses white balance menu options 118 to enter a normal photographic mode. In this mode, white balance is adjusted according to the variable gain coefficients that are enabled, respectively, on variable gain amplifiers 110, 112 and 114 according to a user-selected one of white balance menu options 118. The user is free to obtain other images 138 and process the same using image processing circuits 116. The processing results from other images 138 are stored in normal imaging memory 140. The signal processing circuits 116 may retrieve images from the normal imaging memory 140 to provide image output data 142, e.g., to a display 144.
When the photographer 302 decides to enter field calibration mode 320, this is accomplished by manual interaction with white balance menu options 118, whereupon step 322 entails acquiring and storing a field image 135 of the secondary non-standard target 104. In step 324, signal processing circuits 116 determine FC variable gain coefficients for white balance of the field image 135 to the primary image, e.g., by application of Equations (1), (2) and/or (3). In step 326, signal processing circuits 116 adjust the variable gain coefficients that are enabled on variable gain amplifiers 110, 112 and 114 by replacing the PC coefficients with PA coefficients on each channel R, G and/or B. Photographer 302 may now exit 328 field calibration mode. Camera 100 is next enabled 330 with digital photography calibrated to the “field” secondary image 135 permitting capture of other images 138.
The foregoing discussion is intended to illustrate the features way of example with emphasis upon certain embodiments and instrumentalities. Accordingly, the disclosed embodiments and instrumentalities are not exhaustive of all options or mannerisms for practicing the disclosed principles hereof. The inventors hereby state their intention to rely upon the Doctrine of Equivalents and the following claims in protecting the full scope and spirit hereof.
Number | Name | Date | Kind |
---|---|---|---|
5119178 | Sakata et al. | Jun 1992 | A |
5589879 | Saito et al. | Dec 1996 | A |
6075562 | Sakaguchi et al. | Jun 2000 | A |
20020080245 | Parulski et al. | Jun 2002 | A1 |
20030112342 | Takeuchi | Jun 2003 | A1 |
20040212685 | Smith et al. | Oct 2004 | A1 |
Number | Date | Country | |
---|---|---|---|
20050185065 A1 | Aug 2005 | US |