Colour adjustment

Abstract

A computer-implemented method of adjusting colour output from an electronic visual display. The method comprises: receiving an initial colour to be displayed by a pixel; identifying a shifted colour which reduces an objective function, wherein the colour distance between the shifted colour and initial colour lies within a predetermined range of values; and displaying the shifted colour via the pixel.

Description

[0001]COLOUR ADJUSTMENT The invention relates generally to a colour adjustment method. More particularly, but not exclusively, the invention relates to a method for colour adjustment to reduceunwanted light stimulation from an electronic display.BackgroundVision serves as the primary biological channel for receiving external information. Thehuman eye is the main organ for light perception and discerns electromagnetic radiation with wavelengths from about 380 to 780nm, encompassing imaging and non-image- forming (NIF) vision. Imaging vision processes colour and object shape, while NIF vision, mainly sensitive to light-dark changes, regulates circadian rhythms, which significantly influence cognitive health and behaviour, impacting sleep-wake cycles, mood, alertness, and body temperature regulation. Synthetic light, such as light fromelectronic displays, poses unique challenges to these vision systems, necessitating adeeper understanding and mitigation of its physiological effects.Cone and rod cells in the retina of the eye enable human imaging vision. Cone cells,functional in bright light, differentiate colours via S-cones (sensing blue light with apeak wavelength at 420-440nm), M-cones (sensing green light with a peak wavelengthat 530-540nm), and L-cones (sensing red light with a peak wavelength at 560-580nm).Rod cells, which activate in dim light and focus on light intensity, are most sensitivearound wavelengths of 498nm, within the blue-green spectrum. The NIF vision pathway influences circadian rhythms and is separate from the image- forming pathway. It involves intrinsically photosensitive retinal ganglion cells (ipRGCs) that detect blue-green light with a wavelength around 479nm through melanopsin, influencing physiological activities. ipRGCs are sometimes referred to as “melanopsin photoreceptors”. This pathway’s impact on the human biological clock is significant. Recent advancements in NIF light assessment include metrics like Circadian Illuminance (CIL) and "Melanopic Equivalent Daylight Illuminance" (M-EDI), offering better evaluation of artificial light’s non-visual effects. Despite this, the influence of electronics displays on NIF vision is less examined, with limited research on the effects on melatonin and sleep patterns.In electronic visual displays (e.g., smartphones, tablets, monitors, virtual reality (VR)headsets, etc.) the pixel structure plays a crucial role in shaping the emitted lightspectrum. Predominantly, electronic visual displays utilize RGB (Red, Green, Blue)colour synthesis techniques, which, in contrast to the continuous spectrum of naturaldaylight, tend to produce elevated levels of blue light. Such heightened blue lightemissions are recognized as one of the contributors to visual fatigue. Further, substantialevidence indicates that light from electronic displays can disrupt circadian rhythms,leading to sleeping disorders and hormonal imbalances. This issue assumes even greatersignificance in VR headsets, where the proximity of electronic displays to the eyesintensifies the potential impact.As discussed above, light emitted from the pixels of electronic displays activate the fivekey photoreceptors (S, M, and L cones, rods, and ipRGCs) in the human eye. Two ofthese receptors have sensitivity ranges overlapping with wavelengths of light known to cause adverse effects: one is the S-cone cell, responsible for perceiving short-wave blue light, and the other is the ipRG cells (melanopsin photoreceptors), which induce circadian rhythm changes through perception of blue-green light. Figure 1, which uses a VR headset display as an example of an electronic display, demonstrates the relativenormalised irradiance spectrums of the three emitter types: Blue (Area 1), Green (Area2), and Red (Area 3) in the electronic display, alongside the response curves for humaneye S-cone perception of blue light (Curve 4) and melanopsin response to blue-green light due to the ipRGCs (Curve 5). It is evident from Figure 1 that all blue light and aportion of the green light emitted from the VR headset display overlaps with theseresponse curves. As such, it is desirable to adjust the colour output from electronicdisplays. Although Figure 1 shows the relative normalised irradiance spectrums forthree emitters within a specific VR headset, the normalised irradiance spectrums for other electronic displays are similar, with corresponding peaks at nearby wavelengths.The current approach to alleviate discomfort when using electronic visual displays is toinclude a "Night Mode", which involves applying an orange filter on the screen toreduce blue light intensity. This method leverages the colour adaptation capability of the human eye to maintain relative colour perception. While effective in mitigating light stimulation without hardware modifications, it compresses the original colour gamut, reducing image contrast and negatively affecting object recognition and immersive experience.While "Night Mode" may temporarily be adequate in certain situations (e.g., forgaming) by offering brief visual adjustment, it falls short in professional fields whererigorous colour transmission and extended immersion periods are essential. For instance, the da Vinci Robotic tele-surgery system relies on VR displays for accuratedisplay of human tissues, and earthquake and fire rescue training using VR requiresprecise identification of hazards and vital signs. However, despite significant efforts by developers to create virtual content, existing methods for reducing stimulation have yet to find a balance between colour perception and immersion time. The present invention was devised with the foregoing in mind. Summary of InventionAccording to a first aspect of the invention, there is provide a method of adjusting colouroutput from an electronic visual display. The method may be a computer-implementedmethod.The method may comprise receiving an initial colour to be displayed by a pixel. Themethod may comprise identifying a shifted colour which reduces an objective function, wherein the colour distance between the shifted colour and initial colour lies within apredetermined range of values. The method may comprise displaying the shifted colourvia the pixel. The method may comprise identifying a shifted colour which reduces an objectivefunction subject to the constraint that the colour distance between the shifted colour andinitial colour lies within a predetermined range of values. The shifted colour may be acolour which minimises the objective function subject to the constraint that the colour distance between the shifted colour and initial colour lies within a predetermined range of values. Displaying a shifted colour which reduces an objective function enables the output of a display to be adjusted to reduce unwanted effects. Reducing the objective function subject to the constraint that the colour distance between the initial colour and theshifted colour minimises the perceived change in colour. This is more effective than asimple “night mode”, which reduces blue light without considering colour perception. The initial colour and the shifted colour may be defined using colour space co-ordinates. The initial colour and the shifted colour may be defined using sRGB co-ordinates. The initial colour and the shifted colour may be defined using 24-bit sRGB co-ordinates. The initial colour and the shifted colour may be defined using normalised colour spaceco-ordinates. The initial colour and the shifted colour may be defined using normalisedsRGB co-ordinates.The objective function may be a function of radiant flux from the pixel. The objectivefunction may be a function of spectral radiant flux from the pixel. The objective functionmay be specific to the display. The objective function may be a function of radiant fluxfrom the pixel of the specific display.The objective function may be a function of blue light radiant flux from the pixel. Theobjective function may be a function of radiant flux from the pixel within a givenwavelength range.The objective function may be a function of irradiance received at an observer’s eyefrom the pixel. The objective function may be a function of spectral irradiance receivedat an observer’s eye from the pixel.The objective function may be a function of blue light irradiance received at an observer’s eye from the pixel. The objective function may be a function of irradiancereceived at an observer’s eye from the pixel within a given wavelength range.The objective function may be a function of the activation value of at least onephotoreceptor of an observer’s eye caused by light from the pixel.The at least one photoreceptor may be one or more of: an S cone, an M cone, an L cone, a rod, or an intrinsically photosensitive retinal ganglion cell. The objective function may be a linear combination of the activation values of one or more photoreceptors of the observer’s eye caused by light from the pixel. The objective function may be a function of the activation value of the S-conephotoreceptor of an observer’s eye caused by light from the pixel. The objectivefunction may be a function of the activation value of the ipRGC photoreceptors of an observer’s eye caused by light from the pixel. Reducing an objective function which is a function of S-cone photoreceptor activation may reduce the level of unwanted blue light radiation, which reduces the negative side effects associated with overexposure to blue light. Reducing an objective function which is a function of ipRGC photoreceptor activation may reduce the level of unwanted light which affects the body’s circadian rhythm. Receiving an initial colour may comprise receiving an initial set of colours. Each initial colour in the set of initial colours may be for display on respective pixels in the electronic visual display. The method may comprise, for each initial colour in the set of initial colours, identifying a shifted colour which reduces an objective function. Identifying a shifted colour for each of the initial colours may identify a set of shifted colours. The method may comprise displaying each of the shifted colours. Each shifted colourmay be displayed by the pixel which was to display the corresponding initial colour.The colour distance between the initial colour and the shifted colour may be defined using CIEXYZ coordinates. The colour distance between the initial colour and the shifted colour may be defined using CIELAB coordinates. The CIELAB colour space is designed based on human perception of colour. Defining the colour distance using CIELAB co-ordinates may ensure that the shifted and initial colours are perceptually similar. Defining the colour distance using CIELAB co- ordinates may ensure that the shifted and initial colours are perceptually identical. The colour distance between the initial colour and the shifted colour may be defined using the CIEDE 2000 formula. The colour distance between the shifted colour and initial colour, when defined usingthe CIEDE 2000 formula, may be less than or equal to 5. The colour distance betweenthe shifted colour and initial colour, when defined using the CIEDE 2000 formula, may be less than or equal to 4. The colour distance between the shifted colour and initial colour, when defined using the CIEDE 2000 formula, may be less than or equal to 3. The colour distance between the shifted colour and initial colour, when defined using the CIEDE 2000 formula, may be less than or equal to 2. The colour distance between the shifted colour and initial colour, when defined using the CIEDE 2000 formula, may be less than or equal to 1. Identifying the shifted colour may comprise retrieving a pre-calculated shifted colour associated with the initial colour from a memory. Pre-calculating the shifted colourscorresponding to each of the initial colours and storing them in a memory may reducecomputational complexity as each calculation is only performed once. Identifying shifted colours by retrieving them from a memory may reduce to time taken to identify a shifted colour corresponding to an initial colour. Identifying the shifted colour may comprise calculating the shifted colour. Calculating the shifted colour may comprise identifying a preliminary colour shiftdirection in colour space. The preliminary colour shift direction may be a direction incolour space for which the objective function has a negative gradient at the initialcolour. The preliminary colour shift direction may be the direction in colour space forwhich the objective function has the largest negative gradient at the initial colour. Identifying the shifted colour may comprise identifying an intermediate colour which is offset from the initial colour in colour space along a line defined by the preliminarycolour shift direction. The colour distance between the intermediate colour and theinitial colour may be within a predetermined range of values. Identifying the intermediate colour may comprise using a binary search. The colour distance between the preliminary colour and the initial colour may be defined using CIELAB co-ordinates. Using a preliminary colour shift direction may enable a fast estimate for an intermediate shifted colour which reduces the objective function. Identifying the shifted colour may comprise applying a projected gradient descentalgorithm. Identifying the shifted colour may comprise applying a projected gradientdescent algorithm, starting from the intermediate colour, to identify the shifted colour. The projected descent algorithm may project onto a surface in colour space for which the colour distance from the initial colour is within a tolerance level of a predetermined value. Calculating the shifted colour may comprise inputting the initial colour into apolynomial function. The polynomial function may be stored in a memory. Thepolynomial function may be a fifth-order polynomial function. The polynomial function may be trained via polynomial regression. The coefficients of the polynomial function may be determined via polynomial regression. The coefficients of the polynomial function may be determined via polynomial regression using a set of training input colours and corresponding training shiftedcolours. Each training shifted colour may minimise an objective function subject to theconstraint of having a colour distance relative to the corresponding training initial colour within a predetermined range of values. Using a polynomial function to identify a shifted colour may reduce the computational complexity, and time required, to identify the shifted colour.According to a second aspect of the invention, there is provided a system. The systemmay be configured to perform the method of the first aspect of the invention.The system comprises: a display comprising a plurality of pixels; and a colour processing module configured to: receive an initial colour to be displayed by one of the plurality of pixels of the display; identify a shifted colour which reduces an objective function, wherein the colour distance between the shifted colour and the initial colour lies within a predetermined range of values; and cause the shifted colour to be displayed by said one of the plurality of pixels of the display.The display may be an electronic visual display. The display may be in communicationwith the colour processing module.The system may comprise a memory.The memory may store pre-calculated shifted colours for a plurality of possible initialcolours. The memory may store pre-calculated shifted colours for each possible initialcolour. The memory may store pre-calculated shifted colours for a plurality of possibleinitial colours within sRGB space. The memory may store pre-calculated shifted coloursfor each possible initial colour within sRGB space. The colour processing module may be configured to identify a shifted colour for an initial colour by retrieving the shifted colour from the memory. The memory may store a polynomial function. The colour processing module may be configured to identify a shifted colour for an initial colour by inputting the initial colour into the polynomial function. The colour processing module is configured to calculate the shifted colour. The display may be the display of a smartphone, electronic tablet, television display, virtual reality headset, or augmented reality headset.According to a third aspect of the invention, there is provided a computer-readablestorage medium comprising instructions which, when executed by a computer, cause the computer to carry out the method of the first aspect of the invention. Optional features of any of the above aspects may be combined with the features of any other aspect, in any combination. For example, features described in connection with the method of the first aspect may have corresponding features definable with respect to the system of the second aspect, and vice versa, and these examples are specifically envisaged. Features which are described in the context or separate aspects and examples of the invention may be used together and / or be interchangeable wherever possible. Similarly, where features are, for brevity, described in the context of a single example, those features may also be provided separately or in any suitable sub-combination. Brief description of the drawings The invention will now be described, by way of example only, with reference to the accompanying drawings, in which:Figure 1 shows the relative normalised radiant flux spectra of Red, Green, and Blueemitters within a VR display, together with the spectral response curves for two of the eye’s photoreceptors;Figure 2 shows a schematic view of a system for adjusting colour output from anelectronic visual display;Figure 3 shows a flowchart representing a method for adjusting colour output from anelectronic visual display;Figure 4 shows a flowchart representing a method for identifying a shifted colour;Figure 5 shows a representation in colour space of how a shifted colour is identifiedfrom an initial colour in some examples;Figure 6 shows a representation in colour space of how shifted colours differ from an initial colour;Figure 7 shows the radiant flux as a function of wavelength from a display when usinginitial colours, shifted colours, and a “night mode”;Figure 8 shows the radiant flux as a function of wavelength from for three differentdisplays when using initial colours, shifted colours, and a “night mode”; andFigure 9 shows the colour gamut for three different displays when using initial colours,shifted colours, and a “night mode”. Detailed descriptionFigure 2 shows a schematic view of a system 10 for adjusting colour output from anelectronic visual display. The system 10 comprises a display 12. The display 12 is an electronic visual display. For example, the display 12 can be a virtual reality headset display, an augmented reality headset display, a mobile phone display, a tablet display, a television display, aPC monitor display, or any other electronic visual display. In some examples, thedisplay 12 comprises additional components, such as lenses and mirrors. The display 12 comprises a plurality of pixels. Each pixel in the display 12 is configured to emit light which can be observed by an observer 20. In some embodiments, each pixel in the display 12 comprises a red-light emitter, a green-light emitter, and a blue-light emitter. Each emitter within the pixel can be activated to a specific activation value to provide a specific level of output. For example, in some embodiments, each emitter is an 8-bit light source, meaning there are 256 possible outputs for each light source. In such examples, each pixel is able to display 2563unique colours.The system 10 comprises a colour processing module 14. The colour processing module14 is in communication with the display 12. For example, the colour processing module14 can be electrically connected to the display 12 via conductive wires or tracks, or thecolour processing module 14 can be wirelessly connected to the display 12. The colour processing module 14 is configured to receive an initial colour to be displayed by one of the plurality of pixels of the display 12. The colour processing module 14 can be configured to receive an initial set of colours, each initial colourcorresponding to an associated one of the pixels of the display 12. For example, in someembodiments, the colour processing module 14 is configured to receive ^ colours^^ , ^^ , … , ^^ to be displayed by pixels ^^, … , ^^ respectively.In some embodiments, the initial colour is represented using co-ordinates in an RGB space, such as sRGB space. In some embodiments, the initial colour, or each of the set of initial colours, refers to a set of activation values for red, green, and blue emitters within a pixel. For example, an initial colour may be represented by three activationvalues, such that ^^ = [^^ , ^^ , ^^], wherein ^^ , ^^ and ^^ are activation values for the red,green, and blue emitters of the pixel. In some embodiments, the red, green, and blueemitters are 8-bit emitters, so each activation value is an integer value between 0-255.Such co-ordinates can be referred to as 24-bit RGB colour co-ordinates. References to“colour” herein can refer to the co-ordinates of that colour within a colour space. In some embodiments, colours are defined using normalised activation values between 0-1. For example, in some embodiments, colours are defined via activation values forred, green and blue emitters, wherein each activation value is between 0-1. The use ofnormalised activation values may enable colours to be defined independently of a specific display, which may ensure the methods disclosed herein are usable with all electronic displays, such as HDR standard displays, and potential future displays. The colours used herein can be defined using any suitable colour co-ordinates. The colour processing module 14 can be configured to receive the initial colour, or set of initial colours, from a processor of the system 10. In some embodiments a processor of the system 10 receives an input from an external source (i.e., via an antenna, via the internet, or via a storage device such as a disc or cartridge) and generates an initialcolour for the display 12 based on the input. The processor then sends the initial colourto the colour processing module 14 which receives the initial colour. In someembodiments, the colour processing module 14 is a distinct component from the systemprocessor. In other embodiments, the colour processing module 14 is part of the processor of the system 10. The colour processing module 14 is configured to identify a shifted colour corresponding to the received initial colour. In some embodiments, the colourprocessing module 14 identifies a set of shifted colours, wherein each shifted colourcorresponds to one of the received initial colours within the set of initial colours. Thecolour processing module 14 is configured to communicate the shifted colour to the display 12, such that an associated pixel of the display 12 displays the shifted colour instead of the initial colour. In some embodiments, the system 10 comprises a memory 16. In some suchembodiments, the colour processing module 14 identifies the shifted colour byretrieving the shifted colour from the memory 16. The memory 16 can comprises a look- up-table comprising a set of shifted colours corresponding to a set of possible initialcolours. In other embodiments, the colour processing module 14 identifies a shiftedcolour corresponding to an initial colour by calculating the shifted colour. In some embodiments, the colour processing module 14 performs the method of Figure 3, which is described below. In other embodiments, any component associated with an electronic display 12 can be configured to perform the method of Figure 3. Figure 3 shows a flowchart 100 representing a method for adjusting a colour output from an electronic visual display, such as the display 12 in Figure 2. The method of Figure 3 comprises the steps of: receiving 110 an initial colour; identifying 120 a shifted colour associated with the initial colour; and sending 130 theshifted colour to the display. In some embodiments, this method is performed for aplurality of initial colours, resulting in a plurality of shifted colours being sent to the display. To describe how a shifted colour can be calculated for a given initial colour, theconcepts of colour distance and objective functions are discussed below.Colour distance The smaller the colour distance between two colours, the more similar they are. When the colour distance between two colours is below a threshold value, the two colours become indistinguishable to an observer. The threshold value of colour distance for which two colours become distinguishable is sometimes referred to as the “just noticeable difference (JND)”. The JND depends on the specific viewing conditions and on the function used to define the colour distance.For a given initial colour, the colour processing module 14 can identify a shifted coloursuch that the colour distance between the shifted colour and the initial colour lies withina predetermined range of values. The predetermined ranged of values may be less thanor equal to the JND.In some embodiments, the colour distance is defined using a function which comparesthe co-ordinates of two colours in colour space. For example, for two colours definedusing 24-bit RGB co-ordinates, the colour distance (Δ) between ^^ and ^^ may bedefined as: ∆= ^(^^ − ^^)^ + (^^ − ^^)^ + (^^ − ^^)^where ^^, ^^ , and ^^ represent the activation values between 0-255 for the Red, Green,and Blue emitters respectively in a pixel for ^^.In other embodiments, colour distance is defined using a function which compares theco-ordinates of two colours in an alternative colour space. For example, the colourdistance may be defined using CIELAB co-ordinates, wherein a colour is defined usingan a* (green-red) value, a b* (blue-yellow) value, and an L* (brightness) value.There are various colour distance formulae defined using CIELAB co-ordinates whichare well known to the skilled person. For example, colour distance in CIELAB space can be defined using the CIE76 formula, the CIE94 formula, the CIEDE2000 formula, or any other suitable colour distance formula. As is known to the person skilled in the art, a colour’s co-ordinates in CIELAB space, or any other colour space, can be calculated from the colour’s RGB co-ordinates in RGB space. As such, in embodiments wherein the initial and shifted colours are defined using RGB co-ordinates, the colour distance can still be determined using the equivalent CIELAB co-ordinates. In some embodiments, co-ordinates in sRGB space are first converted to co-ordinates in the CIEXYZ colour space according to a CIE standard, such as CIE1931. Then, the co-ordinates in the CIEXYZ colour space are further converted into co-ordinates in the CIELAB colour space. This process can be implemented in reverse for the calculationof co-ordinates in CIELAB space to co-ordinates in sRGB space.CIELAB colour space is designed based on human perception of colours. Therefore, colour distances calculated using co-ordinates in CIELAB space often provide a better measure of how perceptually similar two colours are. Objective function The colour processing module 14 is configured to identify a shifted colour whichreduces, or minimises, an objective function. The objective function is chosen andreduced / minimised so as to reduce the levels of unwanted / harmful light entering anobserver’s 20 eyes. In different scenarios, different objective functions can be chosen.In some embodiments, the colour processing module 14 is configured to minimise a single objective function. In some embodiments, the colour processing module 14 isconfigured to minimise one of a plurality of user-selectable objective functions. Forexample, a device 10 can comprise a user interface which enables a user to select between a plurality of modes, wherein each mode utilizes a different objective function. In some embodiments, the objective function to be minimised is defined using theharmful / unwanted light emitted from the display 12. In some embodiments, theobjective function to be minimised is defined using the activation of photoreceptors in an observer’s 20 eye.An example objective function, and justification for its derivation, is described below.In other embodiments, alternative objective functions can be minimised.Firstly, it is crucial to understand how colour and light are perceived by the human eye.Consider an image comprising ^ pixels as a set of pairs where each pixel ^^on the screen displays a colour ^^. The irradiance spectrum ^(^; ^^ , ^^), as perceived bythe eye, is a function of the colour ^^ = (^^ , ^^ , ^^) displayed at pixel position ^^, where^^ , ^^ and ^^ are the activation values of the red, green, and blue emitters of the pixel.The irradiance spectrum is influenced by the activation values of red, green, and blueemitters in each pixel ^, represented as measured activation curve or gamma function^(^) = (^⁄ 255 )^.^, and the positional light loss factor ^(^, ^) influenced by the positionof the pixel relative to an observer’s eyes. The relationship is described in the equation: Where, λ denotes the spectrum wavelengths, and ^^(^), in units of ^⁄ ^^ , is the full-power irradiance spectrum at an observer’s eye from each emitter within the pixel. Inmany situations, the positional light loss factor ^(^, ^) can be set as 1.Note that the full-power irradiance spectrum ^^(^) is specific to a given display ordisplay type. Example full-power irradiance spectrums are shown in Figure 1, wherein areas 1, 2, and 3 show the irradiance spectrums for the blue, green, and red emittersrespectively. Therefore, in some embodiments, the irradiance spectrum ^(^; ^, ^) andthe objective function are tailored specifically to a given display.From the irradiance spectrum ^(^; ^, ^), it is possible to determine the associatedstimulation of each photoreceptor of an observer’s 20 eye induced by the irradiancefrom a pixel. ^^(^, ^), the activation of photoreceptor ^ resulting from colour ^ beingdisplayed at pixel ^, can be defined using the equation: Where α represents a photoreceptor, and ^^ symbolizes the α-opic action spectrum forthe associated photoreceptor. Example photoreceptor action spectrums (^^) are shownin Figure 1 – wherein curves 4 and 5 represent the action spectrums for S-cones andipRGCs (melanopsin photoreceptors) respectively. To translate this impact into a standardized measure, the concept of EquivalentDaylight Illuminance (EDI) can be used, specifically referring to α-opic EquivalentDaylight (D65) Illuminance, ^^^,^^^can be computed as per CIE S026 standards [CIE 2018]: Wherein the constants denote the α-opic efficacy of luminous radiation (ELR) for daylight (D65).Having equations for photoreceptor activation ^^(^, ^), it is possible to choose anobjective function to minimize.In some embodiments of the invention, the objective function ^ is a linearcombination of the activation functions of one or more photoreceptors ^. In suchembodiments, the objective function ^ is represented by: In other embodiments, the objective function ^ is a non-linear function of theactivation functions of one or more photoreceptors ^. For example, the objectivefunction L can comprise functions to account for lens effects or other effects. Given that the blue light-emitting units within pixels of electronics displays are designed based on the activation curve of S-cone cells, one assumption that could be used is to equate the issue of blue light stimulation to S-cone cell activation ^^. This assumption allows for an accurate quantification of the irradiance of blue light (short- wave blue light), which activates the S-cone cells while also causing light stimulation. Therefore, in some embodiments of the invention, S-cone cell activation ^^is a keycomponent of the objective function ^.The human body’s need for melatonin fluctuates with external time cues, requiring suppression during the day for alertness and release at night to induce sleep. Ideally,when using electronic displays, the activation of melanopsin (via the IPRG cells) bythe screen should also fluctuate with external time, rather than being fixed or continuously reduced. Therefore, in some embodiments of the invention, melanopsinactivation ^^^^ is a key component of the objective function ^.Given the relative importance of S-cone and melanopsin activation relative to otherphotoreceptors, the activation function ^ is defined in some embodiments as:^ = ^^^^^^^^ + ^^^^ + ⋯Wherein ^^^^ and ^^ are parameters that depend on given situation. For example,telesurgery using a VR headset requires alertness but reduced blue light stimulation,so ^^^^would be greater than ^^.In some embodiments, the objective function ^ is proportional to melanopsinactivation ^^^^, and all ^^ other than ^^^^ are zero. In such embodiments, each shiftedcolour corresponding to an initial colour ^^is defined as ^^^^(^^).In some embodiments, the objective function ^ is proportional to S-cone cellactivation ^^, and all ^^other than ^^are zero. In such embodiments, each shifted colour corresponding to an initial colour ^^is defined as ^^(^^). Although the objective functions are defined herein in terms of photoreceptor activation, alternative objective functions can be defined. For example, the objective function can be defined in terms of irradiance at specific wavelengths or withinspecific wavelength ranges. As research relating to unwanted / harmful electronicdisplay light continues, additional objective functions for minimisation will become apparent. Identification of a shifted colour The purpose of the colour processing module 14 is to perform a mapping function ^()which maps an initial given colour ^ to a new color ^(^) which minimizes an objectivefunction, such as one the objective functions defined above, whilst ensuring that thecolour shift is sufficiently small to be perceptually indistinguishable to an observer 20. Figure 4 shows a flowchart 200 representing a method for identifying a shifted colourcorresponding to an initial colour according to some embodiments of the invention. Insome embodiments, the method of Figure 4 is performed by a colour processing module, such as the colour processing module 14 of Figure 2. To ensure the colour shift is sufficiently small, the colour distance (calculated, forexample, using any of the colour distance formula described above) must be below athreshold value ^. Therefore, the mapping function ^() is subject to the constraint that^(^) ∈ ^^(^), wherein ^^(^) is the set of colours for which the colour distance from theinitial colour ^ is within a specified tolerance level of a threshold value ^.A further constraint of the mapping function ^() is that the shifted colour ^(^) is apossible display colour for the display 12. For example, if the display 12 is configuredto display sRGB colours, then ^(^) ∈ ^^^^.Therefore, in some embodiments, the colour processing module 14 is configured to receive a set of initial colours ^^associated with pixels ^^and identify a set of shiftedcolours ^(^^) obtained using a mapping function ^ which minimizes: subject to the following constraints: ^(^^ , ^(^^) ≤ ^ for ^ = 1, … , ^Wherein ^ is a function defining the colour distance between Methods for solving this optimisation problem are described below. However, the skilled person will recognise that this is a convex optimisation problem and that any suitable alternative method for solving the minimization problem can be used. The colour processing module 14 can identify a shifted colour to solve the optimization problem using a plurality of different methods. In some embodiments, the colour processing module 14 solves the optimisation problem for each initial colour.In some embodiments, the colour processing module 14 retrieves a shifted colour frommemory which is known to solve the optimisation problem for a corresponding initial colour. In some embodiments, the colour processing module 14 applies a formula, such as apolynomial formula, which maps an initial colour to an approximate shifted colour. In such embodiments, the approximate shifted colour may reduce, but not completely minimise the objective function. In some embodiments, the colour processing module 14 receives an initial colour ^^. In some embodiments, the initial colour ^^is defined using RGB co-ordinates. In some embodiments, the initial colour ^^is defined using sRGB co-ordinates. In someembodiments, the initial colour ^^ is defined using RGB co-ordinates and thecorresponding CIELAB co-ordinates are identified.In some embodiments of the invention, the shifted colour ^(^^) is determined by firstdetermining an intermediate colour ^^^^^and then refining the solution, for example,using a gradient descent algorithm.Figure 5 shows a visual representation of how a shifted colour is determined from aninitial colour in some embodiments. The colours in Figure 5 are displayed in CIELABco-ordinates, using L, a, and b axes.For the initial colour ^^ (position 30 in Figure 5), a colour surface ^^(^^) (surface 35in Figure 5) is defined in colour space. ^^(^^) is a surface for which the colourdistance between the initial colour ^^ and every point on the surface ^^(^^) is equal toa threshold value ^. The threshold value ^ can be selected as the maximum colourdistance for which the colour distance is imperceivable to the human eye. For specificapplications where colour preservation is essential, the threshold value ^ will be low.For other applications wherein slight colour change is acceptable, the threshold value^ can be greater.In some embodiments, the colour change is defined using the CIEDE 2000 (ΔE00)formula. In such embodiments, the threshold value ^ can be between 0.1-10. In someembodiments, the threshold value ^ can be between 0.1-6. In some embodiments, thethreshold value ^ can be between 0.1-3. In some embodiments, the threshold value ^can be between 0.1-1.In some embodiments, a preliminary colour shift direction (vector 33 in Figure 5) isidentified. The preliminary colour shift direction is the direction in colour space for which the objective function has the greatest negative gradient when calculated at theposition of the original colour. Each initial colour ^^ has a corresponding preliminarycolour shift direction ^^. For example, for an initial colour ^^, the preliminary colour shift direction ^^can be calculated using the formula: ^^ = −∇^(^^)Having identified the colour surface ^^(^^) and the preliminary colour shift direction^^, the intersection is identified. That is to say, starting from the initial colour ^^, thecolour surface ^^(^^) is reached by moving in the preliminary colour shift direction ^^.The co-ordinates of the intersection are used defined an intermediate colour ^^ ^ (position 32 in Figure 5). In some embodiments, intermediate colour C^^does not lie exactly on the coloursurface B^(C^). For example, if the colour surface B^(C^) is defined using a thresholdvalue ^ = 3, the colour distance between the initial colour C^ ^intermediate colour C^may be equal to 2.99. In some embodiments, an acceptable tolerance value can bedefined using a tolerance. For example, if the acceptable tolerance is set as 1%, thenthe colour distance between the initial colour C^and intermediate colour C^^can beanywhere in the range of 2.97-3.03. In some embodiments, the tolerance is definedsuch that the equivalent 24-bit sRGB values do not change as the colour distance getscloser to the threshold value ^. In some embodiments, the acceptable tolerance is 10%.In some embodiments, the acceptable tolerance is 7.5%. In some embodiments, the acceptable tolerance is 5%. In some embodiments, to determine the intermediate colour ^^^, the co-ordinates are shifted from the initial colour ^^in the preliminary colour shift direction ^^by a fixedstep increment to move away from the initial colour ^^. The colour distance from theinitial colour ^^ is continuously evaluated and this continues until the colour distanceexceeds the threshold value ^. At this stage, the intermediate colour ^^^must liebetween the initial colour ^^ and the “overshoot” colour for which the colour distanceexceeds the threshold value ^, and so a binary search, or any other search, can be usedto locate the intermediate colour ^^^^^. For example, the midpoint between the initial colour ^^and the “overshoot” colour can be located and the colour distance relative to the initial colour ^^can be determined. If the colour distance between the midpoint and the initial colour ^^isless than the threshold value ^, the midpoint is too close to the initial colour ^^ and theco-ordinates are shifted in the preliminary colour shift direction ^^in small incrementsuntil the colour distance exceeds the threshold value ^ again. If the colour differencebetween the midpoint and the initial colour ^^is greater than the threshold value ^, the midpoint is too far from the initial colour ^^and the co-ordinates are shifted in the opposite direction to the preliminary colour shift direction ^^in small increments untilthe colour distance is less than the threshold value ^ again. This process can berepeated, with smaller increment steps each time, until the colour distance relative to the initial colour ^^is within an acceptable tolerance range of the threshold value ^. The intermediate colour ^^^is a colour which reduces the objective function and has an acceptable colour distance from the initial colour ^^, but it does not minimize theobjective function. In some embodiments, to determine the shifted colour ^(^^ ), aprojected gradient descent algorithm is used starting from the intermediate colour ^^^^^ . This is represented in Figure 3, wherein the shifted colours 34, 36 are obtainedby moving along the surface 35 from the intermediate colour. Note that the shiftedcolours 34, 36 are different because they are obtained using melanopsin activation and S-cone activation as objective functions, respectively. For example, in some embodiments, numerical gradient estimation is used to determine the relevant gradients for use in the projected gradient descent algorithm.In some such embodiments, during the optimization, the gradients of ^ and∆^^^^(^, ^) are approximated with small ^ by the following equations: and where ^^is the colour at ^thiteration and ^^is the unit vector along the ^thcoordinate in sRGB space.Using the gradient descent algorithm, the update rule for the colour co-ordinates ^ ateach iteration ^ is given by:^^^^ = ^^ − ^^^where α > 0 is a small step length, and ^^ is the projection of ∇^(^^) onto the tangentplane of the constraint surface ^^(^^) at ^^. The projection is computed as: In some embodiments, the shifted colours ^(^^) are calculated using alternativemethods to the one described above. Any method which minimises the objective function subject to the imposed constraints can be used to identify the shifted colours^(^^). In some embodiments, shifted colours are identified which reduce, withoutnecessarily minimising, the objective function. In some embodiments, polynomial regression is used to determine an approximatecolour shift function ^^^^^^^(). The approximate colour shift function ^^^^^^^ () can beused to identify shifted colours using less computing power. In some embodiments,the approximate colour shift function ^^^^^^^() can be determined based on resultsobtained using a colour shift function ^() which minimises the objective functionusing any of the methods discussed above. For example, a training set of initialcolours ^^^^^^can be used to calculate a set of test shifted colours ^(^^^^^^), and a regression algorithm can be utilised to obtain an approximate colour shift function^^^^^^^ (). The approximate colour shift function ^^^^^^^() can be used subsequentlyto quickly identify shifted colours which reduce the objective function, without necessarily minimising the objective function. In some embodiments, polynomial regression is used to determine an approximatecolour shift function ^^^^^^^() which is a polynomial function. Polynomial regressionherein can refer to regression of a function onto a polynomial basis (for example, the polynomial basis can be Hermite polynomials, Legendre polynomials, Chebyshev Polynomials, etc.).For example, an approximate colour shift function ^^^^^^^ () can be calculated usingknown regression algorithms based on a set of test shifted colours ^(^^^^^^).In some embodiments, the approximate colour shift function ^^^^^^^ () is a fifth-degree polynomial.In some embodiments, the approximate colour shift function ^^^^^^^ () is trained usingtest data to achieve a relative error below 5%.In some embodiments, the approximate colour shift function ^^^^^^^ () is trained usingtest data to achieve an absolute error below 0.5 in RGB co-ordinates. Training theapproximate colour shift function ^^^^^^^ () to have an absolute error below 0.5 inRGB co-ordinates ensures that the shifted colours identified by the approximate colour shift function and identical to the shifted colours identified by the colour shift function using the methods described above. For each shifted colour identified using the approximate colour shift function ^^^^^^^(), the colour distance relative to the initial colour may be calculated to ensure that the colour distance with within a predetermined range of values.As such, for an initial set of colours ^^, the shifted colours ^(^^) can be identified viaseveral methods, including: i. Applying a function to minimise an objective function subject to constraints,as explained above; ii. Applying an approximate function, such as a polynomial function, which istrained using a function that reduces / minimises an objective function subject to constraints; and / or iii. For each initial colour, retrieving a shifted colour stored in memory which mayhave been identified using either of (i) or (ii).The shifted colours ^(^^) reduce / minimise the objective function ^ whilst beingindistinguishable from, or very similar to, the initial colours ^^. Therefore, the present invention is superior when compared with the current solutions of using “night mode”or simply reducing the screen brightness. Figure 6 shows the relative positions inCIELAB colour space of an initial colour 100, a colour 101 obtained by simply reducing screen brightness, a colour 102 obtained using a “night mode” setting, and colours 103, 104 obtained by minimising objective functions. The objective function minimised to obtain colour 103 was proportional to S-cone activation. The objective function minimised to obtain colour 104 was proportional to IPRG cell (melanopsin photoreceptor) activation. As shown in Figure 6, the shifted colours obtained using the systems / methods of the present invention provide entirely different colour changes to the use of “night mode” or the reduction of screen brightness.Figure 7 shows emission spectra for a VR display. Line 50 shows the originalemission spectrum when each pixel ^^ displays the initial colour ^^. Line 52 shows theemission spectrum for the VR display when each pixel ^^displays the shifted colour^(^^) identified using the approach described above. Line 54 shows the emissionspectrum for the VR display when using a traditional “night mode” / orange filter. As is clear from Figure 7, the systems and methods described herein provide similar reductions in unwanted light emissions to traditional “night mode” settings butwithout degrading colour perception. The Blue light peak radiant flux is reduced by35% and the M-EDI value is reduced by 27%.Figures 8(a)-(c) show the irradiance as a function of wavelength from for threedifferent VR headset displays when using initial colours, shifted colours, and a “nightmode”. Figure 8(a) shows the irradiance as a function of wavelength for a HTC Vive Proheadset display when using initial colours 250, shifted colours 252, and a “nightmode” 254. Figure 8(b) shows the irradiance as a function of wavelength for a MetaQuest Pro headset display when using initial colours 350, shifted colours 352, and a“night mode” 354. Figure 8(c) shows the irradiance as a function of wavelength for aPico 4 Pro headset display when using initial colours 450, shifted colours 452, and a “night mode” 454. For each of Figures 8(a)-(c), the shifted colours were identified by minimising an objective function which is proportional to S-cone activation, and by using a colour distance threshold ∆^^^=3.From Figures 8(a)-(c) it is clear that although different displays have differentirradiance spectra, the methods described herein can be applied to consistently reduce the levels of unwanted light.Figure 9 provides further comparison between the display using the initial colours, a“night mode”, and the methods described herein. Specifically, Figures 9(a)-(c) respectively show the colour gamut distributions for an electronic display using the initial colours, a “night mode”, and the shifted colours identified by minimising an objective function.It is clear from Figure 9 that the use of a traditional “night mode” significantlyreduces the colour gamut by reducing the blue colour range, which leads to diminishedcontrast and detail recognition difficulties. In contrast, the colour gamut using themethods of the present invention is substantially similar to the original colour gamut. From reading the present disclosure, other variations and modifications will be apparent to the skilled person. Such variations and modifications may involve equivalent and other features which are already known in the art of electronic displays, and which may be used instead of, or in addition to, features already described herein. Although the appended claims are directed to particular combinations of features, it should be understood that the scope of the disclosure of the present invention also includes any novel feature or any novel combination of features disclosed herein either explicitly or implicitly or any generalisation thereof, whether or not it relates to the same invention as presently claimed in any claim and whether or not it mitigates any or all of the same technical problems as does the present invention. Features which are described in the context of separate examples may also be provided in combination in a single example. Conversely, various features which are, for brevity, described in the context of a single example, may also be provided separately or in any suitable sub-combination. The applicant hereby gives notice that new claims may be formulated to such features and / or combinations of such features during the prosecution of the present application or of any further application derived therefrom. For the sake of completeness, it is also stated that the term "comprising" does not exclude other elements or steps, the term “a” or “an” does not exclude a plurality, and a single processor or other unit may fulfil the functions of several means recited in the claims and any reference signs in the claims shall not be construed as limiting the scope of the claims.

Claims

Claims1. A computer-implemented method of adjusting colour output from an electronicvisual display, the method comprising: receiving an initial colour to be displayed by a pixel;identifying a shifted colour which reduces an objective function, wherein thecolour distance between the shifted colour and initial colour lies within a predeterminedrange of values; and displaying the shifted colour via the pixel.

2. The method of claim 1, wherein the shifted colour minimises the objectivefunction.

3. The method of claim 1 or claim 2, wherein the objective function is a functionof irradiance from the pixel.

4. The method of any preceding claim, wherein the objective function is a functionof the activation value of at least one photoreceptor of the user’s eye caused by lightfrom the pixel.

5. The method of claim 4, wherein the objective function is a function of theactivation value of the S-cone and / or ipRGC photoreceptors caused by light from thepixel.

6. The method of any preceding claim, wherein the colour distance between theinitial colour and the shifted colour is defined using CIELAB coordinates.

7. The method of claim 6, wherein the colour distance between the initial colourand the shifted colour is defined using the CIEDE 2000 formula.

8. The method of any preceding claim, wherein identifying the shifted colourcomprises calculating the shifted colour.

9. The method of claim 8, wherein identifying the shifted colour comprises:identifying a preliminary colour shift direction in colour space, wherein the preliminary colour shift direction is the direction in colour space for which the objective function has the largest negative gradient at the initial colour; and identifying an intermediate colour which is offset from the initial colour in colour space along a line defined by the preliminary colour shift direction, wherein the colour distance between the intermediate colour and the initial colour is within a predetermined range of values.

10. The method of claim 9, wherein the colour distance between the preliminarycolour and the initial colour is defined using CIELAB co-ordinates.

11. The method of claim 9 or claim 10, wherein identifying the intermediate colourcomprises using a binary search.

12. The method of any of claims 9-11, wherein identifying the shifted colourcomprises: applying a projected gradient descent algorithm, starting from the intermediate colour, to identify the shifted colour.

13. The method of claim 8, wherein calculating the shifted colour comprisesinputting the initial colour into a polynomial function.

14. The method of claim 13, wherein the coefficients of the polynomial function aredetermined via regression onto a polynomial basis.

15. The method of any of claims 1-7, wherein identifying the shifted colourcomprises retrieving a pre-calculated shifted colour associated with the initial colour from a memory.

16. The method of claim 15, wherein the pre-calculated shifted colour is calculatedusing the method of any of claims 8-14.

17. The method of any preceding claim, wherein the initial and shifted colours aredefined using sRGB co-ordinates.

18. A system, the system comprising:a display comprising a plurality of pixels; and a colour processing module configured to: receive an initial colour to be displayed by one of the plurality of pixels; identify a shifted colour which reduces an objective function, wherein the colour distance between the shifted colour and the initial colour lies within a predetermined range of values; and cause the shifted colour to be displayed by said one of the plurality of pixels.

19. The system of claim 15, further comprising a memory, wherein the memorystores pre-calculated shifted colours for each possible initial colour; and wherein the colour processing module is configured to identify a shifted colour for an initial colour by retrieving the shifted colour from the memory.

20. The system of claim 15, wherein the colour processing module is configured tocalculate the shifted colour.

21. The system of any of claims 18-20, wherein the display is the display of asmartphone, electronic tablet, television display, virtual reality headset, or augmented reality headset.

22. A computer-readable storage medium comprising instructions which, whenexecuted by a computer, cause the computer to carry out the method of an of claims 1- 17.

Images