WO1996037073A1

WO1996037073A1 - Video signal processing circuit

Info

Publication number: WO1996037073A1
Application number: PCT/US1996/006996
Authority: WO
Inventors: Patrick E. D'luzansky; Victor F. Fleury; James R. Garrett
Original assignee: Polaroid Corporation
Priority date: 1995-05-18
Filing date: 1996-05-16
Publication date: 1996-11-21
Also published as: EP0832533A1

Abstract

A video signal processing circuit for use with an electronic camera. The video signal processing circuit transforms an analog signal representative of an image according to a gamma curve such that the analog signal is elongated along midtones of the image and is contracted at each end of the tone scale. The transformation enhances image quality while the circuit's simple design reduces manufacturing costs. The circuit itself utilizes a resistive tree structure to define cutoff voltages to a clipping circuit where resistive values define the gamma curve.

Description

Title: Video Signal Processing Circuit

Background of the Invention

The present invention relates generally to video signal processing circuits, and, more particularly, the invention relates to video signal processing circuits for applying a specified gamma curve transformation function to a video signal in an electronic camera.

Video signal processing circuits are analog circuits which process video output signals representative of an image from a photosensitive semiconductor device such as a charge coupled device ("CCD"). The video signal processing circuit then applies a transformation function to the video output signal such that a downstream analog-to-digital conversion is optimized to produce better tone reproduction while limiting contouring seen within an output of the image.

The invention has application, by way of example, in an electronic still camera. Such a camera captures an image with a CCD and then converts the analog representation of the image to digital image data representing the image where it may then be processed, stored on an electromagnetic media, or sent to a printing device. A problem with such a camera is that a human eye does not readily perceive tone changes on extreme ends of a tone scale but is extremely adept at perceiving changes in a midrange of the tone scale. In other words, the human eye does not notice changes in black tones or white tones as readily as it perceives changes among the many intervening gray tones. As a result, a linear digitization performed on the analog representation wastes digital image data in the black and white tones and then does not have enough digital image data available to represent the midtones which creates noticeable tone jumps in the image called contouring.

Increasing a byte size of the analog-to-digital converter solves the problem by making more digital image data available in the midtones but substantially increases a cost of the camera.

One commercially available camera uses the analog-to-digital converter itself to transform the image signal. A tap on the analog-to-digital converter which is designed to lower noise is driven with a triangular voltage signal from zero volts to a maximum value and back. The signal creates a piece-wise linear approximation of a Gamma function. A problem with this method is that its approximation is rough having only a limited number of linear steps available to generate the curve.

Summary

The aforementioned and other objects are achieved by the invention which provides a video signal processing circuit for use in an electronic still camera. The video signal processing circuit applies a gamma curve to an analog signal representative of an image to correct the analog signal and to enhance an ability of the electronic still camera to digitize the analog signal. The latter is accomplished by readjusting tone scale mapping such that large changes in tone scale are mapped across a larger digital range while relatively flat changes in tone scale are compressed and mapped to a relatively small digital range. The circuit comprises a cascaded resistance structure, clipping means, buffer means, adder means and current-to- voltage tr.anslation means.

The cascaded resistance structure comprises a resistive tree fed by a reference voltage. The cascaded resistance structure has a plurality of terminals with a resistive element disposed between individual pairs of the terminals such that a resistive value of each resistive elements defines a cutoff voltage at each of the plurality of terminals.

There is a clipping means associated with each of the plurality of terminals and each clipping means is in electrical communication with a current source. The clipping means then draws a current from the current source which is proportional to a voltage of the analog signal. If the voltage of the analog signal becomes equal to the cutoff voltage, the current then becomes substantially constant.

There is a buffer means is associated with each of clipping means to convert the voltage to a current which changes proportionally to the voltage. In the preferred embodiment, buffer means is a transistor which is in saturation such that the currenc into the collector of the transistor is the current and also amplified to boost a magnitude of the current.

The adder means combines the currents from each of the buffer means and forms a single total current.

The current-to-voltage converter means then converts the total current to an output voltage which varies according to said analog signal along the gamma curve. Folowing the fluctuations in the total current, the output voltage varies according to the analog signal but follows the gamma curve which is defined by said cutoff voltages.

In further aspects, the invention provides methods in accord with the apparatus described above. The aforementioned and other aspects of the invention are evident in the drawings and in the description that follows.

Brief Description of the Drawings

The foregoing and other objects of this invention, the various features thereof, as well as the invention itself, may be more fully understood from the following description, when read together with the accompanying drawings in which:

Figure 1 shows a block diagram of an electronic still camera in accordance with the invention;

Figure 2A shows a D-log H curve for the camera shown in Figure 1.

Figure 2B shows a Log V-Log H curve for the camera shown in Figure 1;

Figure 2C shows a graph of scene reflectivity versus print reflectivity for the camera of Figure 1 ;

Figure 3A shows a clipping circuit for use in the video signal processor in the camera according to Figure 1;

Figure 3B illustrates a circuit implementing the gamma curve in a video signal processor for an electronic still camera in accordance with Figure 1; and

Figure 3C shows a voltage output curve out of the video signal processor for the electronic still camera of Figure 1.

Detailed Description

Electronic imaging cameras for recording either motion or still images are in common usage today. Such cameras generally include, as is shown in Figure 1, a two-dimensional photosensitive array which may comprise a high-resolution charge coupled device ("CCD"), charge injection device ("CID"), or other photosensitive sensors. A CCD 16 is depicted in the preferred embodiment but this type of photosensitive array should be considered illustrative and not restrictive. The CCD 16 receives light 12 representative of the image scene in a well-known manner by way of an objective lens and a shutter as shown collectively as optics 14.

The CCD 16 typically comprises a plurality of image sensing elements or pixels arranged in a two-dimensional array with each image sensing pixel converting image defining light reflected from a scene into a corresponding analog voltage value. Sampling is done sequentially for the three primary colors red, green, and blue (hereinafter referred to as "RGB"), and the image sensing elements are preferably arranged in a plurality of rows and columns. For example, in an imaging application the resolution of the electro-optically sampled image comprises approximately 1656 image points, or pixels, per line horizontally and 600 lines vertically. Accordingly, each image has an aggregate 1656 x 600 pixels wherein each pixel is assigned one of the RGB colors.

When an image is captured by an electronic imaging camera, a steady stream of analog voltage values associated with the pixel values for a given row of the image sensing elements are presented to a video signal processor ("VSP") 18 which compresses the analog voltage values that are associated with the highlights of the recorded image while emphasizing the analog voltage values that are associated with the midtones of the recorded image. In essence the video signal processor compresses the tone of the analog voltages by using a tone compression curve having a specified gamma function.

An analog-to-digital converter 20 then transforms row by row the analog voltage values into a plurality of digital electronic image data signals representing the recorded image in a RGB color coordinate system.

The digitized signal is then passed into a processor 22 where it can be stored in a storage device 24 such as an electromagnetic storage device, a hard disk for example, an electro-optical storage device 24, or it may simply be passed onto a computer which is connected to the camera via a cable. In the illustrated embodiment, the signal 26 will be passed externally to the computer.

When the camera 10 is focused upon a subject and a shutter of the optics 14 is opened, varying levels of illumination become incident upon a face of the CCD 16. In shadow areas the level of illumination will be quite low. For highlights, the level will be great. In fact, for every different subject tone there will be a different level of illumination incident upon the CCD 16. These light levels are referred to as illuminants since they deal with light incident upon a film plane, or CCD 16. The camera's shutter speed determines a length of time the illumination representative of an image light will be incident upon the CCD 16.

In conventional photography a negative is exposed to varying amounts of illuminant, H. Upon developing, a resulting image has varying densities according to the exposure. The exposure is the illuminant level multiplied by the exposure time,

E = H x t. Since exposure time, t, is the same for the entire negative, the density of the negative is a function only of the illuminant H. Plotting this function is a D-log H curve. Note also that D = -log T, where T is emulsion transmission. A log - log plot is normally used in photography as a convenient way to express information over a wide range of illumination and transmittances of the negative. Scene reflectance (and hence illuminant levels, H, onto the negative) varies over ranges of 1000 : 1 from highlights to shadows. Expressed in log base 10 this is compressed to a range of 3 : 1.

Illuminants on the CCD 16 generate charge Q, linearly. Charge is linearly converted to voltage, V, by the capacitance of an output amplifier:

Q α H v = δ

C

A D-log H curve as described is shown in Figure 2 A. It can be seen from the figure that the first region 28 has a slope that is very flat and as exposure increases reflection density does not rise noticeably.

At a certain point in the curve the slope increases dramatically. This is shown as the second region 30 where as exposure increases, density increases greatly. In the third region 32, the density again flattens.

In conventional photography, making adjustments to these curves in the photographic material by way of chemistry can adjust how the camera portrays sharpness and contrast in a reflected image.

In the case of an electronic still camera 10 as previously shown, an image coming in that has the log V- log H curve as shown in Figure 2B would be digitized using as many bytes for the flat regions 28, 32 as would be for the second region 30 having a steeper slope. With equal steps between the bytes, large changes in tone between digital steps are introduced as contouring or tone changes in the image which reduce a pleasing effect to the human eye. To avoid contouring, a gamma curve is introduced between CCD 16 and the A-D converter 20 by the video signal processor 18.

The video signal processor 18 transforms an analog signal coming out of the CCD 16 representative of the image such that uniform steps in the A-D are concentrated in the second region 30 of the D-Log H curve. This transform is known as the gamma curve. A designer must define the system tone scale, G, before the gamma can be implemented. G is a tonescale mapping function of a "system." An analog to G in standard photography is the D-log H curve where a psychophysical characteristic of the system is defined by the equation

print ⁼ *J scene

Having chosen the system tonescale map, G, the designer now distributes the limited number of bits in the A/D .along the tonescale so that no contouring will be visible in the display or print. A criteria for an absence of visible contours is that a difference in lightness between adjacent levels should be less than the visual threshold. CIELab is a psychovisual model of uniform color space. L* is a measure of lightness in the CIELab space. Equal increments of L* are perceived as uniform changes in lightness over a wide range of display illuminants and print reflectances. L* can be expressed as a function of display or print reflectance:

where R_Print= print reflectance

R hite = white reflectance

For convenience, R_Whi_te is assumed to be equal to one since the print is printed on white paper. Because CIELab is visually uniform space, equal increments in L* are perceived as equally different. As a result, a well-known criterion for the absence of visual contours is that the changes in L* between levels, ΔL*, should be less than a given value.

Next, a proper step level for L* must be chosen. An example is quantizing the image in steps of equal print L*. With a properly exposed print the quantization steps are then less than 0.4L* apart. This step level is sensitive to an error in exposure such that the quantized steps can become visible. A one stop exposure error will produce differences between levels in excess of 1.5L*.

This step level can be modified to be robust in the presence of the exposure errors that occur. Instead of quantizing in equal steps of print L*, the signal is quantized in equal steps for a wider range of world reflectances than can be printed. This does not produce as uniform steps as with the previous scheme; however, the presence of an exposure error does not produce the large steps seen with the previous scheme.

To find ΔL*, on the print a differentiation is performed:

1 --² ΔL* = - (1 16)(Rprin.) ³ ΔRprin.

This equation states that equal changes in L* are represented by equal changes in the psychophysical perception to the human eye.

To make changes of print reflectance imperceptible, set ΔL* < 0.4 which is below a threshold of visibility.

ΔL* = -(116)(Rpri„.) ³ ΔR _int < 0.4

Rearranging:

1

ΔRprint ≤ — (Rprint) This equation prescribes the maximum reflectance change of the print, ΔR_Pή_nt that is allowed without perceivable contouring.

Given the maximum allowable ΔR_Pπ_nt without contouring and the system tonescale map, G, the maximum allowable ΔR_scene is illustrated in Figure 2C where clearly the ΔR_scene's become smaller in high slope regions of G. Using a derivative of a previous equation, an optimal tonescale map can now be derived:

ΔRscene ⁼ ΔG ΔRpπm

7 = j ΔRscene = j AG ^"'ΔRprintrfR

where γ is the optimal tonescale map. Therefore, for no visible artifacts or contouring to occur, ΔR_pri_nt between quantization steps must be less than ΔG xΔR_scene.

In practice though ΔR_Print= 0 at R_scene = 0, and, therefore, any change would produce an artifact in a linear system. But at R_SCene = 0, R_print (R) = 0 therefore there is no gain. In general, to use a linear analog-to-digital converter ("A D") γ should map scene visibility such that each bit will reproduce an equally perceived lightness change when observing the print:

ΔRpπnt < 0.0072 R£

Implementing the above scheme in an electronic still camera ensures that an image captured by the camera is quantized such that the contours are at least visible in a print printed with a pleasing tone reproduction curve.

Such a gamma curve is implemented in the video signal processor 18. The video signal processor 18 contains circuitry to transform the analog signal representing the image from the CCD 16, referred to as the input voltage V_BM, into a signal as previously described. A subpart of the overall circuit is shown in Figure 3A.

The subpart depicts a clipping circuit which is an integral part of the overall circuit shown in Figure 3B. The clipping circuit utilizes a comparator and a buffer. A current source, i, drives the comparator through a compensating resistor, R_COMP. described in greater detail hereinafter.

The comparator utilizes two branched PNP transistors Ql and Q2 where Q2 is fed by a clipping voltage, V_c, and Ql is fed by an input voltage, VI . The clipping voltage determines the level to which the input voltage, VIN, is compared. The input voltage is representative of an image captured by the CCD. If the voltage VIN is less th.an the clipping voltage, Vc, then current through Q3 will be determined by V_IN R_G2- Once V_I becomes equal to or greater than the clipping voltage, Vc, Ql goes into a cutoff mode and thus no longer effects changes to the current through Q3.

Tr.ansistor Q3 is an NPN transistor whose collector feeds off of an operational amplifier ("op amp"), Ai. The compensating resistor, R_COMP. is then used to match the VBE characteristics of the NPN transistor Q3 to characteristics of the PNP transistors Ql and Q2.

Q3 serves as a buffer, placing V_I across the resistor R_G2- This ensures a current which is representative of the input voltage, VI . This stage, shown as K_\ in Figure 3A, is used to provide voltage-to-current conversion, as well as provide a current gain which is fixed by a ratio of R_GI to R_G2 when VIN < Vc- Voltage-to-current conversion is accomplished by converting a voltage at the base of Q3 into i_c ~ VIN/R_G2- Once V_IN > V_c, I_<- decreases as VI increases, i.e. ΔVI R_sm where

ΔV_IN= V_IN -V_C Thus, a change in gain for VI_N > Vc is a function of R_sm for block Kl.

The op .amp has as a non-inverting input the input voltage, VI_N, increased by 0.7V through a diode, Di. The voltage increase is introduced to level shift V_IN equally with that of the inverting input. In other words, the diode compensates for the voltage increase across the P-N junction of the base-emitter of Q3. Varying the base current to Q3 then varies current drawn from the op amp, Aj, causing the output voltage, VOUT, to vary proportionally.

A smoothing resistor, R_sm, is also shown which is tied to the input voltage, V_IN. The path established between VI_N and the emitter of Q3 ensures that after the cutoff voltage, Vc, is reached by VΠM .and the current drawn from the op amp due to Ki shifts to a constant value, there becomes a current contribution through R_sm producing a smooth transition to the constant value. Graphically, the smoothing rounds the sharp corner produced when the current is suddenly clipped.

However, as the input V_IN increases, a series of other gain blocks, K₂, K₃, et cetera, draw additional current from the op amp depending on the value of VI a d how the clipping voltages, V_c, are defined and, thus, continue to alter the output voltage, V_OUT- When VI is small, all gain blocks draw current from the op amp to produce maximum gain at small values of V_IN- As VIN increases, each gain block stops contributing to an increasing output. The gamma shape is thus defined by gain as a function of the input. In this way the op amp acts like a summing block by adding contributions of the individual gain blocks to produce the gamma curve. The current contributions from each gain block are determined by the clipping voltages and the gains of the gain blocks, thus determining a shape of the gamma curve. The smoothing resistor, R_sm, then acts to provide smoother, sharper transitions on the Gamma curve.

One skilled in the art will realize that the circuit as shown in Figure 3A ignores

DC offsets of the input voltage, VΠ_M, which must eventually be handled in manners well known in the art.

The clipping circuit of Figure 3A is utilized extensively in the video signal processor as shown in Figure 3B. The video signal processor has a tree structure where various taps are utilized along the tree. In describing the figure, exemplary voltage values are used, but these values must be altered for each application to shape the Gamma curve. The tree has a reference voltage, V_R, of one volt. A first resistor Ri having a resistance to effectuate a voltage drop of 0.2V supplies the first clipping circuit with a voltage of Vc₄=0.8V. It should be noted that various increment sizes can be used and a different maximum voltage value can be used without detriment to the invention. A voltage of 0.8V for the first clipping circuit C represents the clipping voltage Vc as previously described. A voltage input being the voltage from the CCD 16 representative of an image is fed into the clipping circuit as V_I and is compared against the clipping voltage Vc =0.8V. The output is a current which runs through the N-P-N transistor Q₃ of gain block -t . This output is then summed and current-to-voltage conversion is performed to provide a contribution to the final output VOUT-

A second clipping circuit is fed off of the tree below R₂ which like before has a

0.2V voltage drop thus providing a clipping voltage Vc₃ = 0.6 volts for the second clipping circuit C₃. VIN is then compared against V_c in C₃ and is again summed to add its contributions to V_out. Likewise the 3rd and 4th clipping circuit compare against Vc₂ = 0.4V and Vci = 0.2V respectively. These currents are all added together with the cumulative signal passing through a current-to-voltage converter to produce . V_out. In the preferred embodiment, the current-to-voltage converter is the op amp Aι_. Figure 3C depicts a typical V_out output which is then passed through an A/D, in this case sampled between 0 and 255, creating a more uniform transition between black and white where contouring has essentially been alleviated by spreading out the steps between 0 and 255 along a more gradual curve.

The invention may be embodied in other specific forms without departing from the spirit or essenti.al characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Claims

Claims .

1. A gamma circuit for use in an electronic still camera for applying a gamma

curve to an analog signal representative of a pixel in an image captured by

the electronic still camera, the gamma circuit comprising

a plurality of piece-wise linear contribution circuits, each piece-wise linear

contribution circuit receives the analog signal and draws a current

according to a voltage magnitude of the analog signal until a fixed

voltage is reached at which current contribution to the gamma circuit

ceases to increase; and

translation means in electrical communication with the plurality of piece- wise linear contribution circuits for converting a total current contribution from the plurality of piece-wise linear contribution circuits into an output signal that substantially follows the gamma

curve.

2. The gamma circuit according to claim 1 wherein the plurality of piece-wise linear contribution circuits each further comprise

clipping means for comparing a voltage of the analog signal to the fixed voltage and generating a second voltage proportional to the voltage of

the analog signal until the voltage of the analog signal reaches the fixed voltage at which point the second voltage becomes substantially constant; and

buffer means in electrical communication with the clipping means for converting the second voltage into said current such that the current changes proportionally to the voltage of the analog signal.

3. The gamma circuit according to claim 2 wherein the clipping circuit further comprises a cascaded resistance tree having a plurality of resistors where each of the plurality of resistors is separated by a node and a voltage at each node defines the fixed voltage for one of the plurality of piece-wise linear contribution circuits.

4. The gamma circuit according to claim 2 wherein the clipping circuit comprises two opposed PNP bipolar junction transistors having as inputs to base sections of the transistors the analog signal and the fixed voltage, respectively.

5. The gamma circuit according to claim 4 wherein the buffer means comprises an NPN bipolar junction transistor where the second voltage is across a compensating resistor in the clipping means which compensates for differences in characteristics between the NPN transistor and the PNP transistors.

6. The gamma circuit according to claim 1 further comprising smoothing means in electrical communication with the analog signal for drawing additional current after the analog signal has reached the fixed voltage such that a transition between contributions from the plurality of piece-wise linear contribution circuits is smoothed.

7. The gamma circuit according to claim 6 wherein smoothing means comprises

a smoothing resistor.

8. The gamma circuit according to claim 1 wherein translation means comprises

an operational amplifier having a feedback resistor, the operational amplifier

receiving the total current contribution from the plurality of piece- wise linear

contribution circuits and converting the total current contribution into the

output signal.

9. The gamma circuit according to claim 8 wherein the buffer means further comprises a gain resistor and the translation means amplifies the output signal with respect to the total current contribution by an amount dictated by the ratio of the gain resistor to the feedback resistor.

10. A video signal processing circuit for use in an electronic still camera to apply a gamma curve to an analog signal representative of an image, said video signal processing circuit comprising

a cascaded resistance structure fed by a reference voltage and having a plurality of terminals with a resistive element disposed between consecutive pairs of said plurality of terminals such that a resistive value of each resistive element defines a cutoff voltage at each of the

plurality of terminals;

clipping means associated with each of the plurality of terminals and in

electrical communication with a current source for developing a voltage proportional to the analog signal until the analog signal

voltage exceeds said cutoff voltage after which the voltage is

substantially constant;

buffer means associated with each of clipping means for converting said

voltage to a current which changes proportionally to voltage;

adder means for combining the current from each of the buffer means to

form a total current; and

current-to-voltage converter means for converting the total current to an output voltage which varies according to said analog signal along the gamma curve.

11. The video signal processing circuit according to claim 10 wherein said buffer means further comprises gain means for amplifying a magnitude of the output voltage.

12. The video signal processing circuit according to claim 11 wherein the current-to-voltage converter is an operational amplifier having a feedback resistor and said gain means includes a gain resistor such that that gain means amplifies the magnitude of the output voltage to a ratio of the feedback resistor to the gain resistor.

13. The video signal processing circuit according to claim 10 wherein said buffer means is a transistor where the current is a collector current.

14. The video signal processing circuit according to claim 10 wherein said resistive elements in the cascaded resistor structure have resistive values which define the gamma curve.

15. The video signal processing circuit according to claim 10 further comprising

a compensating resistive element disposed between said clipping means and said buffer means for matching characteristics of non-linear elements in the clipping means to non-linear elements in the buffer means.

16. The video signal processing circuit according to claim 10 wherein said buffer means further comprises a smoothing resistive element which introduces the analog signal directly into said buffer means so as to smooth a transition in the current once the analog signal has reached the clipping voltage.

17. A method for applying a gamma curve to an analog image signal in an electronic still camera, said method comprising the steps of

comparing the analog image signal to a plurality of clipping voltages and generating a set of second voltages which fluctuate proportionally to the voltage magnitude of the analog signal where the second voltage becomes substantially constant when the voltage magnitude of the analog signal exceeds the clipping voltage;

drawing a current for each of the set of second voltages that is proportional to a magnitude of the second voltages;

adding the current for each of the set of second voltages to form a total current;

converting the total current into a voltage that varies as the analog image signal and substantially follows the gamma curve.