GB2460087A - Modifying pixel values to improve iterative process for holographic display - Google Patents

Abstract

A method of projecting alpha-numeric characters on a screen comprises receiving a frame of alpha-numeric characters (figure 2) defined as intensity varying pixels 101, modifying the pixel values to enhance a subsequent iterative process 102 and modulating a light source to produce an interference signal 103. Thus the alpha-numeric characters can be displayed on a screen and the screen and projector may be placed a long way apart. The iterative process may take the form of a repeated Fourier transform process for a predetermined number of iterations or until a detected error threshold is reached. Modification of the pixels may involve applying a high pass filter to the source image or edge sharpening of the image. The light source may be a laser and the moderator may use phase responsive liquid crystals. Only a part of the display may be used, avoiding using the origin or centre of the display (figure 16).

Description

Projecting Alpha-Numeric Characters

CROSS REFERENCE TO RELATED APPLICATIONS

This application represents the first application for a patent directed toward the invention and the subject matter.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to a method of holography in which alpha-numeric characters or graphiôal images are projected on a screen or into a medium.

2. Description of the Related Art

Projectors are known which project alpha-numeric characters on a screen. Examples of these can be seen used for public display such as at train stations, airports or many other locations. However, limitations exist in terms of the sizes of projectors and screens, and the distances between them in order to acheive an acceptable result. * **

BRIEF SUMMARY OF THE' INVENTION

According to an aspect of the present invention, there is provided a method of holography in which alpha-numeric characters are projected on a screen by the generation of interference patterns, comprising the steps of receiving a frame of alpha-numeric characters defined as intensity varying :. pixels; selectively modifying said pixel values to enhance a subsequent iterative process; and generating an interference signal for the production of said alpha-numeric characters from said modified pixel values.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

Figure 1 shows an overview of procedures according the present invention; Figure 2 shows an example of a frame of characters; Figure 3 shows an example of apparatus suitable for implementation of the present invention; Figure 4 illustrates details of modulator 304; Figure 5 shows an example of differing phase delays; Figure 6 shows a diagrammatic representation of the generation of an interference signal; Figure 7 shows an example of a high pass filter; Figure 8 illustrates higher intensity towards the edges of a character; Figure 9 shows the division of a character into dots; Figure 10 shows an overview of the iterative process in accordance with the present invention; Figure 11 details procedures at step 1001; Figure 12 details procedures at step 1002; Figure 13 shows a method of selective modification of pixel values; Figure 14 shows a first error function calculation method; Figure 15 shows an example of partial reinforcement; Figure 16 illustrates a further example of partial reinforcement; S...

* Figure 17 shows an example of a high pass filter; Figure 18 illustrates the reinforcement of dark areas only; ** * Figure 19 shows an illustration of the modulator programmed in .: 25 accordance with the invention; and Figure 20 shows an example of a screen with alpha-numeric characters projected onto it.

DESCRIPTION OF THE BEST MODE FOR CARRYING OUT THE

INVENTION

Figure 1 An overview of procedures according to the present invention is shown in Figure 1. At step 101, a frame of characters is received. The characters are alpha-numeric, and may include letters, numbers, logos etc. An example of this is given in Figure 2.

At step 102, the pixel values received at step 101 defined as intensity varying pixels are selectively modified in order to enhance a subsequent io iterative process. As will be described with reference to Figures 3 to 6, in order to produce a frame of characters such as that shown in Figure 2, an interference signal is generated at step 103 from the modified pixel values produced at step 102.

The result of application of the generated interference signal is the modulation of the phase of light received from a light source, such that the representation of the desired alpha-numeric characters is projected onto a screen.

Figure 2 * ** An example of a frame 200 of characters is shown in Figure 2. In this example, three lines of text are shown at 201, 202 and 203. In this example, *.* the text relates to train departure times. In order to produce this desired frame * ** of characters on a screen (as shown in Figure 20) an interference signal is * calculated such that a light source has its phase modulated appropriately to produce an image such as that shown in Figure 2 on a screen.

*. 25 The alpha-numeric characters are to be projected as light onto a dark

background 204, as represented by the shading.

Figure 3 An example of apparatus suitable for implementation of the present invention is shown in Figure 3. Apparatus 301 includes a receiver 302, a processor 303, a modulator 304, a light source 305 and lenses 306 and 307.

Receiver 302 is configured to receive a frame of alpha-numeric characters defined as intensity varying pixels. This information is passed to processor 303 which is configured to selectively modify the pixel values to enhance the subsequent iterative process; the process that generates an interference signal for the production of the alpha-numeric characters from the modified pixel values. The interference signal is supplied to modulator 304, which is configured to receive the interference signal and modulate a light source (in this example laser 305) as a result. The alpha-numeric characters are therefore projected as an interference pattern. In this example, light source 305 is a laser which passes through lenses 306 and 307 before reaching modulator 304. Lens 306 is a diverging lens and lens 307 is a converging lens.

Modulator 304 is in this example a spatial light modulator of the type illustrated in Figure 4 which has an effect as illustrated in Figure 5. Thus, light leaves laser 305, passes through lenses 306 and 307, reflects from modulator 304 and leaves the apparatus at 308.

As an example, a laser light source may be collimated to provide a beam of typically 2mm across. The beam then needs to be spread out. S...

Therefore a diverging lens is followed by a converging lens placed at the appropriate focal length to produce a parallel beam. The beam is required to S..

focus on the screen. The screen is in the far field, which may be considered a 25 long way off such that any distance from a point on the screen to any point on the lenses is essentiaiiy the same distance. An examp!e of a distance considered to be at the far field is a distance greater than ten times the size of the lens. For example, a lens of about 10mm across could be used thus meaning that further than 1 metre away is far field. In practice, the apparatus may be used to project tens of metres or even further so that the screen itself

is well into the far field.

The apparatus seeks to modulate the phase of different parts of the beam. This is achieved by placing a piece of material in the way in which the refractive index can be changed on a pixel-by-pixel basis. This is further described with reference to Figures 4 and 5. The use of a device including mirrors (as shown in Figure 4) does create a mechanical issue in that the light source must be far enough away so that the beam does not clip the lens on the way out of the apparatus. Thus, the box will need to be relatively long. In contrast, in alternative embodiments devices where the light beam passes through the modulator rather than being reflected by mirrors would be used.

In an alternative embodiment, an additional lens may be placed at the exit of the apparatus (at 308). This would enable the size of the screen used to be increased.

As an example, pixel spacing may be around 6 microns which is five times the wavelength of visible light. If pixels are close together there is a wider deflection angle. Consequently, a twenty foot picture at twenty metres is possible. In alternative configurations different ratios and sizes may be employed. Appropriate lenses which could be used to increase the display size are available as are used as zoom lenses for projectors.

As an example, a light source such as a laser capable of emitting two to *0* five Watts would be appropriate for a public display producing an image two metres by one metre. The distance from the projector to the screen could be anything from around ten metres to around a one hundred metres. Thus, a great degree of flexibility is provided in terms of the sizes and locations of the projector and the screen relative to one another.

Figure 4 Details of modulator 304 is shown in Figure 4. In this example, the light mod ulating device takes the form of an array of phase responsive liquid crystals. Three crystals are shown in this diagram. A first crystal 401 is shown next to a second crystal 402 and a third crystal 403. In the present embodiment, a large array of liquid crystals is provided. For example, a square array of 2000 by 2000 liquid crystal devices could be used. A transparent layer is shown at 404 which allows light to pass through it and protects the array of liquid crystals.

Each liquid crystal has associated with it a reflective surface such as surface 405 shown for liquid crystal 401. In this example each pixel is a mirror topped with a liquid crystal. A top coating of glass (404) has a conductive layer of tin oxide connected to one electrode with the mirror connected to the other.

A transistor connects the mirror to one of two voltages to achieve the switching operation between one of two stable states.

Each of the array of crystals has the property that when a voltage is applied across it, the properties of the liquid crystal change. In this example, * the property that changes is that the amount of phase delay introduced by the liquid crystal is altered. The degree of phase delay is dependent on the voltage **** level applied across the liquid crystal. The voltage that is applied to each liquid crystal can be altered individually. A silicon chip backplate is provided at 406 onto which the liquid crystals are mounted. Thus a liquid crystal on silicon (LCoS) apparatus is provided. A back plate is also provided at 407. A voltage *. 25 can be applied between reflective layer 405 and transparent layer 404 in order to affect the properties of the liquid crystals. Silicon plate 406 transmits the voltages to the reflective plates.

Light enters the apparatus of Figure 4 at, for example, point 408. The light passes through transparent layer 404, passes through liquid crystal 401 and reflects from reflective surface 405. The light then passes back through liquid crystal 401, out through transparent layer 404 and leaves the apparatus at point 409. In this example, a coherent light source such as a laser is used therefore at point 408 the light waves are all in phase. After passing through the liquid crystal 401 the light waves may or may not have had their phase altered depending upon the voltage applied to crystal 401. A second beam of io light is seen entering the apparatus at point 410. The light entering the apparatus is in phase with the light entering at 408. This light passes through transparent layer 404, liquid crystal 402, reflects from the reflective back plate and leaves the apparatus at 411. At point 411 this second beam of light has had its phase modulated by a different degree from the first beam of light. This is because the two liquid crystals 401 and 402 have had different voltages placed across them. Thus, the beams of light leaving the apparatus at 409 and 411 are no longer in phase. A diagrammatic representation of this is provided in Figure 5.

In an alternative embodiment, a piezo-electric crystal is also utilised in order to further alter the light. A piezo-electric crystal allows the liquid crystals to be moved by a tiny amount, this can be used to apply dither which reduces *S..

* the appearance of noise and/or speckle, or for other applications. ****

Figure 5 An example of the differing phase delays is shown in Figure 5. The first light beam shown entering the apparatus at 408 and the second light beam entering the apparatus at 410 are seen to be in phase at 501. After passing through liquid crystals 401 and 402 respectively it can be seen at 502 that the beams leaving the apparatus at 409 and 411 have been modulated to alter their phase to different deglees.

When the apparatus shown in Figure 4 is scaled up to have a large number of liquid crystals, the light which is emitted (after having its phase altered) produces an interference pattern. This pattern forms the display of alpha-numeric characters on a screen.

Figure 6 A diagrammatic representation of the process undertaken in order to io generate an interference signal for the production of an alpha-numeric character is shown in Figure 6. For simplicity, a single character is shown in this diagram. It should be appreciated that in fact the entire image is processed as a whole, characters are not treated individually. Therefore, for display of the image as shown in Figure 20 the whole of the frame shown in Figure 2 would be processed.

A representation of an alpha-numeric character (in this case the letter L) shown at 601. At 602 the modulator is represented and at 603 the projected image of the character at 601 is shown. Light is seen to enter at 604 and reflect from modulator 602 in order to produce the illumination of character 603 *:* 20 as illustrated by arrow 605. Thus, the goal is to generate an interference signal such that modulator 602 modulates light source 604 in such a way as to S.,. produce alpha-numeric character 603 on the screen. *.S

* Further arrows are shown at 606 and 607. Arrow 606 represents the process undertaken to go from a frame of alpha-numeric characters (601) to a modulator (602). This process takes the form of a forward Fourier transform.

Arrow 607 represents a transformation from modulator 602 to image 603, which in practice is an inverse Fourier transform. By assessing the quality of the image produced at 603 (as will be further described with reference to Figures 13 and 14) an iterative process may be undertaken such that the interference signal supplied to modulator 602 is improved with each iteration until either a given number of iterations have been completed or a sufficiently good result is achieved at 603. The iterative process is undertaken as shown in Figure 10 and further described in Figures 11 and 12.

The modulator panel is in the Fourier domain and the screen is in the intensity domain. In a forward direction the intensity is used and the phase is io left out because the eye cannot see it therefore we set it to zero. Going back to the Fourier domain the phase and the amplitude are obtained. The amplitude is uniform across the display because a uniform laser is being shone at it.

Therefore, variations in amplitude are zeroed. The process is repeated, each time zeroing out everything that is not required at each end. The phase representation gradually improves resulting in a better and better image.

Thus, the algorithm starts with the letters and then goes through an iterative process to calculate the phase pattern on the grid of pixels. This is an iterative Fourier transform. A frame is transformed by taking in a pair of intensity and phase values (real and imaginary) and performing a Fourier transform to give another complex pair which is converted back to intensity and phase by an inverse Fourier transform. The intensity is the intensity of light passing through and phase is the phase of that light. It is not possible to modify intensity so the intensity is set to unity. The phase is left at its return value and an inverse transform is performed upon this pair to give a complex 25 pair back at the start. Thus, the inverse Fourier transform yields the intensity in phase that would be obtained if we were to project through the modulator uniform laser light with the intensity set at unity but the phase manipulated. In use, the inverse Fourier transform is performed in the air which is what the far field may be considered to do. When done once, there is a lot of light spilling off. This needs to be improved so the algorithm is repeated. The new intensity and new phase values are taken from the inverse transform. The phase is set to a constant number and the intensity values are kept. Each time the process is undertaken a better result is achieved.

If multicoloured displays are required, different wavelengths require different lenses, so it is necessary to provide three sets of lenses in the apparatus shown in Figure 3 in order to focus three coloured beams onto the surface of the light modulator 304. Alternatively it is possible to use a diffractive optical element to apply focusing. This element can be mathematically combined with the diffraction pattern on the LCoS device. By calculating the phase response of a suitable lens, perfect focus can be achieved for different wavelengths without the use of expensive acromatic physical lenses. The use of cheaper spherical optics can then be enhanced by applying an aspheric correction to the virtual lens on the panel. The lens phase response is in the form of spherical patterns of equal phase. These can then be calculated algorithmically in FPGA hardware to allow real-time adjustment. For two axis (X and Y) we calculate the phase contribution using an arithmetic series for * ** 20 example; for each pixel horizontally PX(n) = PX(n-1) + dX(n), PX(n) donates phase for pixel X at location n. dX is similarly calculated: dX(n) = dX(n-1) + ddX(n) and ddX is similarly calculated: ddX(n) = ddX(n-1) + dddX. dddX is a constant for a given lens profile and initial values of dX, ddX and dddX are also *e.

* set for a given lens profile. This maths is performed by modulo 2Pi so the 25 phase contributions wrap. The Y contribution is calculated in the same manner and added, again, to the modulo 2P1. This firmware focus mechanism can be used to compensate for poorer optics, curved screens, differing wavelengths of light and the creation of an image in space away from the screen.

The device is required to direct photons to the letters/numbers of the display under display and away from the surrounding area. It is therefore necessary to calculate an interference pattern in the form of a grid of phase delays. The degree to which a phase delay is introduced will depend upon the wavelength of the light therefore it is necessary to calibrate the device for each wavelength.

In a four-colour system, the colours can be multiplexed. For example it is possible to achieve switching at 200Hz. Therefore four or five colours could be included. This would result in better colour balances compared with just using red, green and blue.

If only a small amount of text, for example a single word or letter is to be displayed then all the energy is being directed towards that text. Therefore the text would tend to be very bright. It is therefore necessary to control the laser source in order to provide compensation. For example current control the laser source in order to provide compensation. For example current control could be provided at frame rate. The algorithm starts with the alpha-numeric characters to be displayed and goes to an iterative process to calculate the phase pattern on a grid of pixels. This is an iterative Fourier transform. This process is further described with reference toFigures 10 to 12. * ** * **.

Figure 7 In order to improve the efficiency of the iterative process, one possibility is to modify the source image. An example of a high pass filter is shown in * Figure 7 as a graph 701 with an undershoot 702 an overshoot 703. The results of the application for high pass filters such as that shown in Figure 7 is shown in Figure 8. The effect is to push the energy to the edge of each character.

This is further described with reference to Figure 8. Starting with a high pass filtered version of the required image requires fewer iterations of the process for any given amount of black.

Figure 8 In Figure 8 an extract of the display of Figure 2 is shown, in this example a letter L shown at 801. A representation of Figure 8 is an intensity image and the text is shown as having edges with higher intensity and with lower intensity in the middle of the letter. In an example, the filtering is tuned to give typically an 80% to 60% reduction in the middle of each letter which in turn results in more energy being directed towards the edges. If this approach is not adopted, light tends to creep around the edges of the characters which is seen as a pattern in the form of bright dots. Thus as a result of high pass filtering the energy is pushed to the edges of the characters. This has a number of advantages including making the algorithm converge more quickly and producing an output of higher quality. Whilst brightness in the middle of the character is sacrificed, this does not tend to be perceived as an issue, as brightening of the edges of the characters enhances the appearance of sharpness.

* As part of the application of the high pass filter, the application of the undershoot 702 results in negative intensities which whilst physically S...

S..' meaningless are kept in the calcu'ations for the purposes of the algorithm.

Thus, in Figure 8 it can be seen that there are more light dots such as 802, 803, 804 etc towards the edge of image 801 compared with fewer light *: : dots such as a 805, 806 etc towards the middle of the letter. When the :. 25 algorithm is undertaken iterative!y, light is gradually pushed towards the edges resulting in a sharpening of the image.

Figure 9 An example of a further enhancement to improve the algorithm is shown in Figure 9. In this example, the fill of a character is cut up into dots.

Figure 9 shows character 901 divided notionally into a large number of pixels such as 902, 903, 904 etc. Pixels may be grouped into dots, each dot being a group of pixels. For example, pixels 902, 903, 904 and 905 may be grouped into one dot as represented by bold outline 906. Once cut up into dots, the energy can be put into the dots. Thus as far as the eye is concerned reconstruction of the letter shape takes place. A more efficient clustering of io dots is therefore achieved. It has been found that the best gains are achieved by modifying the image source rather than the algorithm itself.

Figure 10 The iterative process carried out in accordance with the present invention is outlined in Figure 10. Process 1 is shown at 101 which includes a forward Fourier transform, and is further detailed in Figure 11. Process 2 is shown at 1002 which includes an inverse Fourier transform and is further detailed in Figure 12. The result of process I is that the final phase is set and the amplitude is averaged. As a result of process 2 final amplitude is set and * partial reinforcement takes place. After process 1 and process 2 have been carried out, a question is asked at 1003 as to whether the process has *** finished. This may be decided either by a predetermined number of iterations or by a decision being made as to whether or not the result is sufficiently good.

If the question asked at 1003 is answered in the negative indicating that the process is not finished then control passes back to 1001. Procedures :. 25 terminate when the question asked at 1003 is answered in the affirmative.

Figure 11 Procedures carried out at step 1001 are further detailed in Figure 11. At 1101 and 1102, real and imaginary components of the image (after processing such as high pass filtering, clustering etc if these procedures have taken place) are used as inputs into a Fourier transform which takes place at 1103. The Fourier transform at 1103 produces outputs at 1104 and 1105 which are again real and imaginary components. Both real components 1104 and imaginary component 1105 are then used as inputs to two functions. A first function at 1106 calculates the phase and a second at 1107 calculates the amplitude. The io result of the phase calculation at 1106 is a matrix as shown at 1108. The result of amplitude function shown at 1107 is a matrix as shown at 1109. In this embodiment, the amplitude is then averaged as shown at 1110 to produce a new matrix at 1111. The amplitude is not required at this stage therefore setting it to its own average has been found to be advantageous A further matrix called FinalA is, at step 1112 set to equal matrix A calculated at step 1108.

Matrix Al at 1108 and matrix B2 at 1111 are inputs to a conversion function at 1113. This function converts the matrices back into a real component at 1114 and an imaginary component at 1115. * *S

* S * -Figure 12 S...

Process 2 shown at 1002 is further detailed in Figure 12. The real and imaginary numbers output from process I are shown as inputs to process 2, at 1114 and 1115.

An inverse Fourier transform is undertaken at 1201, resulting in a new real and imaginary component as shown at 1202 and 1203 respectively.

These new real and imaginary components are fed into two functions. Firstly they go into an amplitude function shown at 1204 then secondly a phase function shown at 1205. The result of the amplitude function is the production of a matrix B3 as shown at 1205. This matrix B3 is fed into a function such that it is saved as FinaiB at 1206, and is also fed into a function to partially reinforce the amplitude at 1207. The partial reinforce function also requires the input matrix [0] as shown at 1208. This matrix represents the original image data. Further details as to the nature of the partial reinforce function at 1207 will be described with reference to Figures 15 to 18. The result of partial reinforce is the production of a new matrix B4 as shown at 1208. This matrix B4, along with a matrix A2 shown at 1209 produced by the phase function are inputs to a conversion function shown at 1210. This function converts the data in matrices B4 and A2 into a new real and imaginary component as shown at 1211 and 1212 respectively.

Figure 13 In order to improve the situation with the iterative function it is necessary to calculate how close the result is and produce an error function.

A method of selective modification of pixel values is shown in Figure 13.

At 1301 a first interference signal is created and at 1302 the first interference * signal is subtracted from the ideal image in order to produce an error function.

At 1303 the error function is removed from the first image.

Figure 14 A first error function calculation method is shown in Figure 14. The method as previously described is iterated at 1401. At 1402 the intensity vakies calculated are summed. At step 1403 the differences from the original image are summed in order to produce an error function.

As the projector is used for alpha-numeric characters, the text itself would not change very often therefore it is possible to pre-calculate values.

Values represent the image as a whole as distinct from calculating values for individual characters. The Fast Fourier Transform (FFT) works well for this application and can be implemented in software. Implementation in gate arrays make even better results possible. However, it is appreciated that updates in real time may be required therefore alternative embodiments provide real time processing. This also provides functionality for example which enables the inclusion of advertising which may include a moving video element.

The process can be initiated with random phase values. It has been found that a higher quality image can be achieved if the random values are low pass filtered before use.

Additionally, intensity values may be forced to somewhere between the original picture and unity, which effectively produces a de-saturated image.

This results in a faster convergence but the algorithm is made more complex.

Figure 15 An example of partial reinforcement is shown in Figure 15. The image space 1501 is shown to be defined about an origin 1502. When the partial * reinforce function is carried out at 1207 as described in Figure 12 it is possible to reinforce values for only the shaded area 1503. Thus, in this example, only the top two quadrants of the display area 1501 are reinforced. In alternative * *a embodiments alternative configurations of partial reinforcements could be *** * used.

*: .: It has been found advantageous to reinforce only some of the display . 25 area. An example of an advantage is that any image projected in the top left quadrant for example will produce a shadow in the bottom right quadrant. Thus the use of only the top quadrants removes this issue.

The biggest cause of light spreading where you do not want light is a limitation of the panel (modulator) itself. Liquid crystal that is applied onto the mirrors of the panel has a tendency to mix up adjacent mirrors, such that when different phases are applied to adjacent mirrors the result that is achieved is a mixture of the two. This is a limitation of spatial frequency response in the Fourier domain. As a result you tend to get texture in the background and the letters.

In the intensity plane, if you do nothing you get zero in the centre. The io object is to move the spot away from the centre (origin). The further it is moved, the more frequency response you require across the width and height.

Therefore more and more detail is obtained in the Fourier plane as you move away from the origin. There is therefore an advantage in trying not to move away from the origin too much which is a further advantage in only working in two quadrants.

Figure 16 A further example of partial reinforcement is shown in Figure 16. In this example the origin 1502 is intentionally excluded from the area to be reinforced 1601. There is a tendency for light to always leak out at the origin therefore avoiding the origin has been found to be advantageous. *...

: .. Figure 17 A previously discussed, a limitation of the modulator is that the liquid crystal is imperfect and tends to blur signals to adjacent pixels. The further a beam is steered from the origin the more noise is generated in phase response. Therefore the signal given to the panel may be altered in order to compensate for the inadequacy of the panel. The liquid crystal molecules are large molecules that do not change characteristics rapidly. There is therefore an advantage to altering the signal so as to cause the molecules to move further in the direction in which they are required to move. This can be achieved by applying a high pass filter to the two-dimensional phase image before putting numbers into the panel.

An example of a high pass filter similar to that described with reference to Figure 7 is shown in Figure 17. In this case a high pass filter is to be applied to the interference signal which has been generated as described with reference to previous Figures. Filter 1701 has an undershoot 1702 and an io overshoot 1703. Filter 1701 is applied after the iterative process previously described, It has been found that this produces a better result in terms of the end projected image; The characteristics in liquid crystals tend to be different horizontally compared to vertically. Therefore a different frequency response is tuned to horizontally compared to vertically. This enables us to apply a different frequency response for the horizontal and vertical, as the high pass filter is generally done horizontally then vertically. We may for example need more boost in one direction than the other, such as requiring more boost of a high frequency in the long axis of a rectangular display.

* 20 By placing a photodiode at the origin we are able to measure the :::::: success. The dimmer the dot at the origin, the more efficient the process is. It is therefore possible to tweak the high pass filter cut off point to minimise the amount of energy at the origin.

I

*:*. Figure 18 A further refinement to the process which can be used is shown in Figure 18. This example is reinforcing only the black (dark) areas and not the light areas. An extract of the image is shown at 1801 and at 1802 an example of an image produced part way through the iterative process is shown. It can be seen in this image that an approximation to the original image has been produced, with a number of light areas where dark areas should be and vice versa. At 1803 an example of the image with the blacks reinforced but the whites not reinforced is shown. Thus the black areas are forced towards zero but the light areas of characters are untouched. We are therefore sacrificing resolution in the characters but as a result get a greater contrast ratio. This means that the process requires far fewer iterations, for example it may require 80 iterations instead of 150. In some instances abetter result is also achieved, particularly in large areas of illumination for example a logo. It has been found.

that this is perceived as sharper to the human eye for given number iterations than an image where both blacks and lights have been reinforced. Thus, the partial reinforce function at 1207 is adapted accordingly.

Figure 19 An example of the result of the iterative process as previously described is shown in Figure 19. The modulator 304 is provided with a grid of Os and Is for example 0 at 1901 and 1 at 1902. Thus the modulator modulates the * phase of light which hits any pixel coded by a I and does not modulate the phase of light hitting a pixel coded by 0. This occurs by adjusting the voltage to **** each individual cell as described with reference to Figure 4. Thus, as a result of shining a laser at a modulator configured shown in Figure 19 an interference pattern is produced which is shone onto a screen and gives a result as shown *:*. in Figure 20. S.

Figure 20 An example of the end result is shown in Figure 20. A large screen 2001 is shown projected with image 2002 which is representative of the original desired image as shown in Figure 2. * S. * . * * ** *.** * I * * * S S... *.*

I I. S * SS * ** * I..

S

Claims (20)

1. Claims 1. A method of holography in which alpha-numeric characters are projected on a screen by the generation of interference patterns, comprising the steps of: receiving a frame of alpha-numeric characters defined as intensity varying pixels; selectively modifying said pixel values to enhance a subsequent iterative process; and io generating an interference signal for the production of said alpha-numeric characters from said modified pixel values.
2. 2. A method according to claim 1, wherein said pixel values are selectively modified by a process of high pass filtering.
3. 3. A method according to claim I or claim 2, wherein said pixel values are selectively modified by making intensity variations within the body of each character.
4. 4. A method according to claim 3, wherein said intensity variations define a plurality of dots.
5. 5. A method according to any of claims 1 to 4, wherein the selective modification of the pixel values is performed by: II'S 25 creating a first interference signa' based on an original image; subtracting the first nterferer1ce signa! from the origina! image to produce an error function; and removing the error function from the original image.
6. 6. A method according to any of claims I to 5, wherein a first error function is calculated by: summing the intensity values after a first iteration to calculate interference values; and summing all the differences from the original image to produce an error function.
7. 7. A method according to any of claims I to 67 wherein said generating step performs an iterative process.
8. 8. A method according to claim 7, in which said intensity values are forced to somewhere between an original picture and unity.
9. 9. A method according to claim 7, wherein said generating step includes reinforcing said interference signal.
10. 10. A method according to claim 9, wherein said reinforcement only occurs in areas of interest of a display area. e.'. * . *
11. 11. A method according to claim 9, wherein said reinforcement is applied only to dark areas of an image. **

    *. 25
12. 12. A method according to any preceding claim, wherein said . interference sIgnal is generated for only part of an avai!able display area.
13. 13. A method according to claim 12, wherein said interference signal is not generated for the origin of the display area.
14. 14. A method according to any preceding claim wherein said interference signal is selectively modified by a process of high pass filtering.
15. 15. A method according to claim 7, wherein an input to said iterative process is a set of phase values.

    10.
16. 16. A method according to claim 15, wherein said phase values are random.
17. 17. A method according to claim 15 or claim 16, wherein said phase values are selectively modified by a process of low pass filtering.
18. 18. Holography apparatus for the projection of alpha-numeric characters onto a screen, comprising: a receiver configured the receive a frame of alpha-numeric characters defined as intensity varying pixels; a processor configured to selectively modify said pixel values to enhance a subsequent iterative process, and generate an interference signal * .. for the production of said alpha-numeric characters from said modified pixel I..* values; 0S* * a modulator configured to receive said interference signal and modulate a light source as a result, thus projecting said a'pha-numeric characters as an **.* interference pattern.
19. 19. Apparatus according to claim 18, wherein said light source is a laser.
20. 20. Apparatus according to claim 18, wherein said modulator uses liquid crystals to modulate the phase of said light source. * *S ** . * S. a. * S **** * .. *. S S...S S.. ** *SSSS S..S