S&F Ref: 880157 AUSTRALIA PATENTS ACT 1990 COMPLETE SPECIFICATION FOR A STANDARD PATENT Name and Address Canon Kabushiki Kaisha, of 30-2, Shimomaruko 3 of Applicant: chome, Ohta-ku, Tokyo, 146, Japan Actual Inventor(s): Kieran Gerard Larkin, Stephen James Hardy Address for Service: Spruson & Ferguson St Martins Tower Level 35 31 Market Street Sydney NSW 2000 (CCN 3710000177) Invention Title: Colour Calibration and Measurement Method The following statement is a full description of this invention, including the best method of performing it known to me/us: 5845c(1830441 1) - 1 COLOUR CALIBRATION AND MEASUREMENT METHOD FIELD OF INVENTION The current invention relates to colour calibration and the measurement of colour. DESCRIPTION OF BACKGROUND ART 5 Accurate colour for imaging and printing is a fundamental requirement for producing visually pleasing results. Satisfying individual preferences for colour requires making adjustments to the colour balance and distribution of a scene that are not faithful to the original colours of the scene. However, these adjustments can often be described as systematic changes from a photometric colour space, so having an accurate base from 10 which to make these adjustments is crucial to the quality of the final results. Accurate colour reproduction based on input devices requires calibration of the colour response of the input device. Examples of such input devices are scanners and cameras. Each type of device requires a different method of calibration. Scanners and cameras can be colour calibrated by taking a colour test chart with 15 many patches of known spectral reflectance and making a scan or photo of the chart. A software system then analyses the image of the chart and the known spectral reflectances to determine a mapping between the output of the scanner or camera and a well-defined colour space, such as CIE LAB colour space. A colour produced by a later scan or photo is related to this known colour space by using the relationship between the output values 20 of the image device to the values on the calibrated chart to infer the corresponding colour in the well-defined colour space. Alternately scanners and cameras may have their sensors calibrated directly. This may be done by exposing the sensor to a known amount of narrow bandwidth light and 1826184_1 880157_speci -2 measuring the counts produced by the sensor. For instance, a camera may be positioned so that it is imaging the exit aperture of an integrating sphere. The integrating sphere is attached to a monochromator and a spectrophotometer. The monochromator allows the selection of the wavelength of the light that exits the integrating sphere and also its 5 bandwidth. The spectrophotometer measures the output from the integrating sphere directly as a function of wavelength. The camera may be used to image the exit aperture of the sphere for a selection of wavelengths covering the visible band from 380nm to 730nm. The number of counts registered on the camera sensor for each wavelength can then be compared to the intensity of the light measured by the spectrophotometer to 10 determine the spectral sensitivity of the camera. This information can then be used to derive colour measurements from subsequent images captured by the camera. This sensor measurement system is extremely expensive, time consuming and difficult to establish. Each calibration of a sensor requires multiple images across the visible wavelength band and is therefore also very time consuming. 15 For printing devices, accurate colour reproduction is generally achieved by printing a colour test target consisting of many patches on the printing device to be measured and then measuring the reflectivities of the patches using a scanning spectrophotometer. Generally many hundreds of patches must be measured which means that the measurement of this chart can often take many minutes and generally 20 requires an expensive measurement apparatus. SUMMARY OF THE INVENTION The current invention substantially ameliorates these deficiencies in the prior art methods of colour calibration of imaging and printing devices. In accordance with one aspect of the present disclosure, there is provided 1826184_1 880157_speci -3 In accordance with another aspect of the present disclosure there is provided BRIEF DESCRIPTION OF THE DRAWINGS 5 At least one embodiment of the present invention will now be described with reference to the drawings, in which: Fig. I illustrates a direct vision prism; Fig. 2 illustrates the refractive indices of the glass materials used; Fig. 3 is a illustration of the trajectory of a light ray through the direct vision 10 prism; Fig. 4 is an illustration of the angular deviation of a perpendicular ray incident on the direct vision prism; Fig. 5 is an illustration of the suggested geometrical configuration of the first embodiment; 15 Fig. 6 is a block diagram for the functioning of the first embodiment; Fig. 7 is a block diagram of the focal length calibration step of the first embodiment; Fig. 8 is a block diagram of the spectral sensitivity measurement step of the first embodiment; 20 Fig. 9 is an illustration of the suggested geometrical configuration of the second embodiment; Fig. 10 is an illustration of the elements that comprise the test chart configuration region; Fig. 11 is a block diagram for the functioning of the second embodiment; 1826184_1 880157_speci -4 Fig. 12 is a block diagram of the focal length calibration step of the second embodiment; and Fig. 13 is a block diagram of the patch reflectance measurement step of the second embodiment. 5 DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION The invention will be described in with reference to two embodiments. The first embodiment describes a system for measuring the spectral response of a sensor. The second embodiment describes a system for measuring the spectral reflectances of a printed test chart with many test patches. 10 Before the embodiments are described, a background section describing light deviation in passing through a direct view prism will be presented. Mathematical background A direct view prism is a refractive element that is formed of at least two prisms, typically a pair of prisms, cemented together. The prisms are cemented along their 15 hypotenuse to produce a block with parallel faces. The prisms are made from materials with the same refractive index nd but different Abbe number Vd. An example of a direct view prism is shown in Figure 1. It consists of a prism made from H-LaK7 glass cemented to a prism made from ZF3 glass. The dispersive properties of the materials with wavelength were modeled using 20 the Sellmeier equation (1). B,__ ___ B22 B2 n2(A)=I+ + + (3 18-C,84 8-C2 2 -C 1826184_1 8801 57_speci - 5 Figure 2 shows the refractive index for the glasses used in Figure 1. These refractive indices cross at 699nm. At this wavelength, light crosses the interface between the prisms without deflection. Figure 3 shows a diagram representative of a ray entering the direct vision prism 5 at an angle 0 with respect to the normal of the prism surface. The ray leaves the prism with angle 0+5, with deviation due to the prism denoted 3. The refractive index of the first prism is n,, the refractive index of the second prism is n 2 , and the refractive index of air is no z1. The prism angle is a and we have internal refraction angles # and y. Applying Snell's law at the interfaces leads to the following equations: nosin 0=n, sin#I 10 n, sin(a+/8)= n 2 sin y n 2 sin(y- a)= no sin(9+ 6) The internal refraction angles # and y can be eliminated from this equation, leading to an equation that relates the input and output angles to the variables that describe the prism properties: nosin~cosa+ nf -2 sin 2 Osina = n sin(+5)cos a+ n 2 n2 sin 2(9+ b)sin a 15 This equation does not have a simple solution for 5 in terms of the other variables, but it may be solved numerically. In the limit of small deviation the solution may be written sinasec6(n' + -n, -n sin 2 69n -n sin 2 9+n sin 2 ) no(no sin sina -cosa n> -n 2sin29) In the further limit that the angle of incidence is 0, we have that 20 5 z n, - n2 tana n0 1826184_1 880157_speci -6 The deviation as a function of wavelength of the prism shown in Figure I is shown in Figure 4. In this figure the deviation is measured in degrees. First embodiment The first embodiment of the invention consists of an apparatus used to measure 5 the spectral sensitivity of an imaging sensor. The configuration for this measurement apparatus is shown in Fig. 5. The apparatus includes a calibrated light source, 510, a diffuser plate, 520, a slit, 530, a direct vision prism, 540, a camera to be calibrated, 550, with a connection to a computer, 560. The computer is not shown. The calibrated light source can be of two 10 different types, either a lamp with emission lines at known wavelengths, or a lamp with a known irradiance as a function of wavelength. As seen in Fig. 5, the prism 540 is disposed between the slit 530 and the camera 550. The diffuser plate 520 and slit 530 form a mask by which the light source 510 can impinge on the camera via the prism 540. The functioning of this apparatus will now be described in more detail with 15 reference to Fig. 6. In the first step, 610, the camera, 550, is positioned spatially such that the slit, 530, is central to its field of view. The slit, 530, the front face of the prism, 540 and focal plane of the camera, 550, are approximately parallel, the slit is oriented so that the image of the slit on the camera sensor is vertical, and the prism is oriented so that the dispersion is oriented horizontally on the camera sensor. 20 In the second step, 620, the relationship between the angular deviation of the prism as a function of wavelength and the position of the dispersed light on the sensor plane of the camera is determined. In the third step, 630, this information is used to make a measurement of the spectral sensitivity of the camera. 1826184_1 880157_speci -7 The calibration of the relationship between the angular deviation of the prism as a function of wavelength and the position of the dispersed light on the sensor plane of the camera is now described in more detail with reference to Fig. 7. First, in step 710, a light source with known emission lines in its output spectrum is used as the calibrated 5 light source. In the preferred embodiment the light is a Mercury/Argon lamp. In step 720, an image of the slit is captured by the camera, and in step 730 is downloaded to the computer. In step 740 the region of the image downloaded to the computer that contains the image of the dispersed slit is cropped from the larger image. This data is summed vertically in step 750. Due to the emission lines in the lamp spectra, this summed data 10 will consist of a signal with many peaks corresponding to the emission lines of the lamp. These peaks are detected in step 760 and the positions of these peaks are used to determine the focal multiplier in step 770. Step 770 is achieved by determining the value of the focal length F that minimises the difference between predicted positions of the emission lines having passed through the prism and the actual positions detected in 15 step 760. This relationship specifies the correspondence between pixels in the projected image and wavelength. The measurement of the spectral sensitivity of the camera performed in step 630 is now described in more detail with reference to Fig. 8. In step 810 an incandescent source with known irradiance as a function of wavelength is used as the calibrated light 20 source 510. An image of the slit is captured in step 820 and in step 830 is downloaded to the computer. In step 840 the region of the image downloaded to the computer that contains the image of the dispersed slit is cropped from the larger image. This data is summed vertically in step 850. In step 860 the intensity of each pixel in a vertical strip is determined by dividing the summed pixels by the number of rows that have been 1826184_1 880157_speci -8 summed. In step 870 the ratio of the intensity calculated for each pixel in step 860 is divided by the calibrated output of the lamp at the wavelength determined for this pixel in step 760. This ratio is the spectral sensitivity of the pixel to this wavelength of light. As an alternative to the above processing procedure, the slit 530, may be replaced 5 with a mask with a two dimensional noise pattern. Rather than projecting the image data in steps 750 and 850, a deconvolution step can be performed to deconvolve the noise pattern from the dispersed image to recover the spatial distribution of the light as a function of wavelength. This approach may be superior to that described due to improved signal to noise. However, the choice of noise patterns to be used is critical in 10 ensuring an accurate result. Noise patterns such as cyclical shifted M-Sequences laid out in a two dimensional array with periodic boundary conditions are the preferred embodiment in this configuration. Second embodiment As a second embodiment of the invention, an apparatus to measure the spectral 15 reflectances of a printed test chart, is now described with reference to Fig. 9. A camera, 910, is positioned to image a test chart region, 930, through a direct vision prism, 920. The test chart region is illuminated by light sources, 940, and light from these sources is prevented from directly entering the camera using baffles, 950. The printed test chart represents a spectral colour reproduction of the printing device from which the chart was 20 formed. The test chart region, 930, is described in more detail with reference to Fig. 10. In this figure element 1010 represents an empty test chart region. Element 1020 is a cover that is placed over the test chart region that consists of a black region with long slits excised, 1050, and another black region, 1060, which has a white stripe of known 1826184_1 880157_speci -9 spectral reflectance along its middle. Element 1030 is a representative colour chart that can be placed on the test chart region, and element 1040 depicts the test chart region with the cover, 1020, placed on top of the test chart, 1030. The functioning of this apparatus will now be described in more detail with 5 reference to Fig. 11. In the first step, 1110, the camera, 550, is positioned spatially such that the test chart region, 930, is central to its field of view. The test chart region, 930, the front face of the prism, 920 and focal plane of the camera, 910, are approximately parallel. The camera is oriented so that the image of the slits on the test chart cover, 1020, on the camera sensor is vertical, and the prism is oriented so that the dispersion is 10 oriented horizontally on the camera sensor. In the second step, 1120, the relationship between the angular deviation of the prism as a function of wavelength and the position of the dispersed light on the sensor plane of the camera is determined. In the third step, 1130, this information is used to make a measurement of the reflectivity of each patch on the test chart, 1030. 15 The calibration of the relationship between the angular deviation of the prism as a function of wavelength and the position of the dispersed light on the sensor plane of the camera is now described in more detail with reference to Fig. 12. First, in step 1210, a light source with known emission lines in its output spectrum is used as the calibrated light source. In the preferred embodiment the light is a Mercury/Argon lamp. In step 20 1220, the camera captures an image of the test chart region, and in step 1230 the image is downloaded to the computer. In step 1240 the region of the image downloaded to the computer that contains the image of the dispersed calibration region 1060 is cropped from the larger image. This data is summed vertically in step 1250. Due to the emission lines in the lamp spectra, this summed data will consist of a signal with many peaks 1826184_1 880157_speci -10 corresponding to the emission lines of the lamp. These peaks are detected in step 1260 and the positions of these peaks are used to determine the focal multiplier in step 1270. Step 1270 is achieved by determining the value of the focal length F that minimises the difference between predicted positions of the emission lines having passed through the 5 prism and the actual positions detected in step 1260. This relationship specifies the correspondence between pixels in the projected image and wavelength. The measurement of the reflectivity of the patches on the test chart performed in step 1130 is now described in more detail with reference to Fig. 13. In step 1310 an incandescent source is used as the light source 940. An image of the test chart region is 10 captured in step 1320 and in step 1330 is downloaded to the computer. In step 1340 a loop over all the patches in the test chart region, including the calibration patch, is entered. In step 1350 the region of the image downloaded to the computer that contains the image of the dispersed slit for the current patch is cropped from the larger image. This data is summed vertically in step 1360. In step 1370 the intensity of each pixel in a 15 vertical strip is determined by dividing the summed pixels by the number of rows that have been summed. If there are more patches to be analysed then processing is returned to step 1350 to operate on the next patch. After all patches have been analysed to determine the intensities of the dispersed slits for each patch, in step 1370 the ratio of the intensity calculated for each test chart 20 patch is divided by the intensity measured for the calibrated patch. This ratio is the reflectivity of the patch at the wavelength determined for this pixel in step 1260. As an alternative to the above processing procedure, the slits in the mask 1020, may be replaced with a two dimensional noise pattern. Rather than projecting the image data in steps 1250 and 1360, a deconvolution step can be performed to deconvolve noise 1826184_1 880157_speci pattern from the dispersed image to recover the spatial distribution of the light as a function of wavelength. Variations In the two embodiments of the invention described above, many specific choices 5 were described which are not central to the functioning of the invention. In particular, the imaging device used was a camera and it would be equally possible to use another imaging device, such as a scanner sensor, microscope, or video camera. Other calibrated light sources may be used, such as LEDs or Arc lamps etc. Alternately, the output of the illuminating lamps may be measured by a calibrated spectrometer as each image is taken 10 by the camera and the output of the spectrometer could be used as the reference for calculating the spectral sensitivity of the imaging device or the reflectance of the patches. Finally, the embodiments above describe the calibration step as being performed for each measurement. This is generally unnecessary as it may required to be done only irregularly or even only once for each apparatus or type of apparatus. 15 The above descriptions of the two embodiments of the invention have been made with reference to a monochrome camera. The invention is not restricted to monochrome cameras and may be applied to colour cameras. For cameras where colour is sampled using a Bayer pattern, then each of the colour channels may be independently analysed by treating each colour channel as an independent array of colours. For cameras which 20 are configured to capture 3 colour channels simultaneously, then each colour channel may be analysed independently. The forgoing described only some embodiments of the present invention and modifications can be made thereto without departing from the spirit and scope of the present disclosure. 1826184_1 880157_speci