Derivation of electromagnetic spectrum in multi-stimulus color values for analog video cameras

Abstract

In rendering an image of a virtual scene, color information of at least one surface element of the virtual scene is saved in the form of color space coordinates of a multi-stimulus color space. The surface element is configured with a number of basis functions which describe a relative wavelength distribution of light components emitted by the surface element, respectively, and which model a spectral radiance density as a linear combination of the basis functions. Proportionality factors of the basis functions are derived by calculating the spectral radiance density with at least one color space coordinate of the color information from the surface element and at least one color matching function configured to the color space coordinate, and the spectral radiance density is calculated by substituting the proportionality factors. The synthetic camera data is calculated by processing the spectral radiance density in a camera simulation of a camera arranged in the virtual scene.

Description

Technical Field

[0001] Cameras are increasingly used in intelligent automation systems, which can derive detailed images of their environment from camera footage and react independently to events within that environment. Such systems can be, for example, highly automated, even autonomous, vehicles or robots that can move independently and reliably in unknown environments. Another example is a surveillance system that can independently identify dangerous or unwanted situations in camera images and issue alerts. Background Technology

[0002] Such camera-based systems often perform safety-critical tasks and require thorough evaluation before mass production. Evaluation in real-world environments is typically time-consuming because critical events are rare and therefore unreproducible. Therefore, it is common practice to conduct test runs of the system, at least partially, in a virtual environment where camera data is synthetically generated based on the virtual environment and provided to the system under test.

[0003] When the camera-based system is based on trained artificial intelligence (AI), such as AI for object recognition, facial recognition, or scene interpretation, similar problems arise with AI training. Definite training scenarios may be few and unreproducible. Therefore, there is an increasing trend towards training AI in virtual environments using synthetic camera data.

[0004] Important not only for evaluating but also for training camera-based systems, the synthetic camera data must be credible, meaning it possesses the highest possible similarity to real camera data. This allows results obtained in virtual environments using synthetic camera data to be applied to the real world. For this, it may be necessary to realistically simulate the technical physical processes in a camera, processes that are involved in generating camera data in a real camera, especially when synthetic camera data is used to mimic low-processing camera data, such as raw camera data generated by an image sensor.

[0005] The light emitted from a real environment that illuminates a real camera image sensor has a spectrum. This light consists of a continuous spectrum of different wavelengths. Typical graphics engines used to render virtual environments (such as Unreal Engine, Unity Engine, and CryEngine) typically display colors using the RGB color space as a mixture of a small number of discrete color values. This is a tristimulus color space, which displays each hue as a mixture of three discrete color components: red, green, and blue. While this display achieves a realistic color impression to the human eye, it does not correspond to physical reality. Even assuming perfect simulation of the image sensor and imaging optics, a camera simulation fed with RGB color values ​​does not produce raw camera data corresponding to real cameras.

[0006] To simulate light phenomena, especially the exposure and image data generated by digital cameras, so-called spectral rendering is known in the prior art. Here, color is modeled as a spectrum rather than discrete multi-stimulus color values ​​from the outset. Examples are the professional paper "Digital Camera Simulation" by Joyce Farrel, Peter Catrysse, and Bran Wandell (Applied Optics 51(4), 2012) and patent publication US 5,710,876A. However, the computational cost for this is empirically so high that real-time camera simulation with sufficient quality is impossible.

[0007] To reduce computational costs, the spectrum is evaluated using given RGB values, thus utilizing the already existing RGB pipeline for spectral light simulation. Such a scheme is disclosed, for example, in the professional paper "GPU Rendering of the Thin Film on Paints with Full Spectrum" by Roman Ďurikovič and Ryou Kimura (Tenth International Conference on Information Visualisation, 2006). This paper proposes replacing the given RGB values ​​with a linear combination of three predetermined spectra that produce the color impressions of red, green, or blue in the human eye. While this method is fast and implementable, it is a coarse approximation and unsuitable for simulating infrared or ultraviolet cameras operating outside the visible spectrum.

[0008] Methods for calculating synthetic camera data for different designs based on camera simulation and virtual scenes are also known to those skilled in the art. For example, they can be found in the paper “Full Spectrum Camera Simulation for Reliable Virtual Development and Validation of ADAS and Automated Driving Application” by René Molenaar et al. (IEEE Intelligent Vehicles Symposium (IV), 2015). Summary of the Invention

[0009] In this context, the objective of the present invention is to improve the rapid evaluation of a spectrum composed of predetermined multi-stimulus color values.

[0010] This invention relates to a computer-implemented method for generating synthetic camera data. The method includes: rendering an image of a virtual scene and storing the color information of at least one face element of the virtual scene in the form of color space coordinates in a multi-stimulus color space. The face element may be, for example, a polygon, a surface element, a texture, a pixel, or a side face of a volume pixel. A multi-stimulus color space is understood as a color space that models color as an N-dimensional vector or tuple of values, where each vector coordinate represents a discrete, predetermined color value. Computer graphics, especially those of general-purpose graphics engines, are mostly based on a three-dimensional RGB color space, such as the sRGB color space.

[0011] Light emitted from a surface element first follows the formula

[0012] (1)

[0013] Modeled as a linear combination of fundamental functions B of quantity m. L is the spectral radiance density of the surface elements, and each fundamental function is the relative wavelength distribution of the light components emitted by the surface elements. Therefore, the spectral radiance density is described by m electromagnetic spectra with proportionality factors c1, ..., c1 to be determined. m The superposition of.

[0014] The scaling factor for the linear combination is derived by calculating the spectral radiance density using color space coordinates from at least one color information source of the face element and at least one color matching function configured for the color space coordinates. For this purpose, the following formula can be used for each color space coordinate:

[0015] (2)

[0016] It is a color space vector The coordinates. In the sRGB color space, for example... Vector Includes color matching functions configured for color space coordinates. In the sRGB color space, It is a linear transformation of the CIE-1931-RGB color matching function. It is a constant specific to the color space. Equation (2) is originally used to assign a set of color space coordinates k to a given electromagnetic spectrum described by spectral radiance L, which gives the human eye the same color impression as the given spectrum. In the method according to the invention, the color space coordinates of the surface element are known. Substituting equation (1) into equation (2) yields the system of equations.

[0017]

[0018] (3)

[0019]

[0020] To be used to determine the scaling factor The proportional factor Substituting into equation (1) yields the spectral radiance density of the surface element.

[0021] The synthesized camera data is calculated by processing the spectral radiance density in a camera simulation designed for this purpose, which is set up in the virtual scene.

[0022] To derive the electromagnetic spectrum, the method calculates physically plausible pre-assumptions by combining the radiation spectrum of the surface elements (Equation 1) with the multi-stimulus color values ​​of the surface elements predetermined by a virtual scene. The spectrum derived in this way is closer to reality than that of methods known in the prior art and thus produces synthesized camera data of better quality, while also being fast enough for real-time camera simulations. This approach is necessary when the synthesized camera data is set up to stimulate a test piece designed as independent hardware located outside the computer system used to generate the synthesized camera data according to the invention (“Hardware in the Loop”). Such a test piece waits for current camera data at predetermined time intervals, and the method accordingly must calculate and provide the camera data within the same time intervals. The repetitive execution of the method steps required for this purpose is advantageously combined with scene simulation, which simulates the temporal development of a virtual scene, such as the movement of traffic participants in a road traffic scene. The scene is imaged, and the result is also synthesized camera data, which is then used to image the instantaneous state of the virtual scene in each execution, wherein the instantaneous state includes, in particular, the instantaneous position, spatial orientation, or pose of objects in the virtual scene.

[0023] Solving the integral requires the maximum computational cost to derive the scale factor according to equation (3). However, since the integral is neither location-dependent nor time-dependent, it only needs to be calculated once. For each face element configured with one or more basic functions, the integral to be solved for deriving the scale factor can be pre-calculated, and scene simulation can begin after the pre-calculation is complete. A further step to determine the scale factor is applied after solving the integral in equation (3). The computational cost is so small that it can be implemented in real time without any problems.

[0024] When the surface element is designed as a passive radiator, it is advantageous that at least the first fundamental function... (Without limitation on generality) Consider the spectral profile of the illumination of the surface element and the reflectance spectrum of the surface element. The spectral profile of the illumination can optionally be designed as a standard light type, such as standard light type A (incandescent lamp), D65 (standard fluorescent lamp), F1 to F6 (fluorescent tubes, various types), and LED-RGB1 (white LED). The reflectance spectrum can be configured, in particular, to material properties, which can then be configured or assigned to the surface element. Software tools for creating virtual scenes may, for example, include a material database with multiple different material properties that can be configured to different surface elements in the virtual scene. A material-specific reflectance spectrum is configured for each material property in the database, and the separately configured reflectance spectra are automatically assigned to the surface element by configuring the material properties to the surface element.

[0025] Preferably, the color space coordinates are calculated considering the spatial orientation of the surface element relative to the simulated camera, that is, considering the angle between the camera's optical axis and the surface normal of the surface element. Such angular correlation is known in academia as a bidirectional reflection distribution function. Its consideration is originally embedded in many graphics engines when calculating the color space coordinates of surface elements. The use of angle-dependent color space coordinates within the scope of the method according to the invention produces angle-dependent spectral simulations, which in the prior art can only be performed with high computational cost and non-real-time.

[0026] The method is also suitable for simulating infrared or ultraviolet cameras that operate entirely or partially outside the visible spectrum, i.e., below 380 nm or above 780 nm. For this purpose, the spectral radiance is calculated such that it includes desired wavelengths outside the visible spectrum, for example by appropriately selecting the integral boundary in equation (3). .

[0027] The camera data calculated using the method described above is preferably designed as raw camera data. Raw camera data is understood as unprocessed data directly output from the camera's image sensor. Normally, raw camera data is a matrix of electrical parameters (voltage, current intensity, or charge), each reflecting the exposure of an image sensor pixel. The electrical response of each image sensor pixel is derived from the spectrum of light exposed to the pixel and the pixel's quantum efficiency. Quantum efficiency is the wavelength-dependent sensitivity of a pixel. Considering at least one quantum efficiency of the simulated image sensor of the simulated camera, for example, a quantum efficiency uniformly applicable to all simulated sensor pixels, the raw camera data can therefore be directly calculated from the calculated spectral density or from multiple spectral densities calculated using the method according to the invention.

[0028] Preferably, the synthesized camera data is read in and processed by at least one component of an electronic system designed for processing the camera data. In this case, an interface for providing the camera data may be provided, allowing the at least one component to read in the camera data. The component or electronic system may be a prototype to evaluate the prototype's response to the camera data. Instead of evaluation, the synthesized camera data may also be configured to train the electronic system, particularly a neural network of the electronic system. The electronic system may, in particular, be an electronic system under development. The component may include a complete electronic system or any substructure of the electronic system, such as a subsystem, structural assembly, electronic component, component, single processor, single FPGA (Field Programmable Gate Array), or similar component, software, or software component for executing hardwired program logic.

[0029] Based on its adaptability to simulating infrared cameras, the method is particularly suitable for evaluating or training assistance systems configured to evaluate camera data from a camera pointed at a person for monitoring their status. The person is, in particular, someone who can operate a vehicle or machine. Such assistance systems often operate in the infrared range. Thus, the assistance system functions even in darkness, or possibly illuminates the person with infrared light without glare. The virtual scene in this case includes a virtual representation of the person, and the imaging of the virtual scene correspondingly includes an image of the person. The spectral profile of the illumination can correspondingly mimic the infrared (or perhaps also ultraviolet) illumination of the person performing the operation.

[0030] The present invention also relates to a computer program product including means for implementing the method according to the invention. The present invention also relates to a computer system including means for calculating synthesized camera data according to the method according to the invention, and further including an interface for providing the synthesized camera data for reading and processing by at least one component of an electronic system designed for processing the camera data. Attached Figure Description

[0031] The accompanying drawings and the following description illustrate embodiments of the present invention. As shown:

[0032] Figure 1 This shows the data channels used to generate camera data;

[0033] Figure 2 This illustrates the angle-dependent generation of multi-stimulus image data; and

[0034] Figure 3 An exemplary virtual scene is shown; Detailed Implementation

[0035] As in Figure 1 The computer system shown in the diagram typically includes three components: scene simulation 2, camera simulation 4, and test piece 28.

[0036] Scene simulation includes rendering 6 of scene 42. For this rendering 6, a graphics engine is utilized, such as Unreal Engine from various vendors available on the market, for example, from Epic Games, Inc. Scene 42 is a 3D scene populated with multiple static and moving graphical objects. The design of scene 42 depends on test piece 28 and testing requirements. In principle, scene 42 should include aspects important to the specified utilization of test piece 28, which is set up to identify and evaluate these aspects. If test piece 28 is, for example, a part of a highly automated vehicle pedestrian recognition system, then scene 42 should be designed to include a road traffic scene with virtual pedestrians. If test piece 28 is a virtual surveillance camera used for automatically identifying drowning accidents, then scene 42 should display waters including virtual swimmers, showing some typical behaviors of swimmers in distress.

[0037] Figure 2 The illustration clarifies the steps of rendering 6. The graphics engine renders scene 42 using the sRGB color space, which optimizes the display of scene 42 on the monitor for a realistic appearance to the human eye. Scene 42 consists of multiple textured polygons 34, which form the basic face elements of the surfaces visible in scene 42. The graphics engine renders scene 42 from the perspective of a virtual camera 32 and saves color information in the form of RGB vectors 12 for each face element 34 visible to the virtual camera 32. The RGB vectors 12 include three color space coordinates for the color values ​​of red, green, and blue. However, the values ​​of these three coordinates depend on the spatial orientation of the face element 34 relative to the virtual camera 32, i.e., on the solid angle between the straight line 40 connecting the virtual camera 32 to the face element 34 and the face normal 38 of the face element 34. For this reason, a correlation with the position of the light source appears in the face elements 34 that are reflecting light. In the figures, this correlation is represented by different amounts of RGB vectors 12 along different spatial directions. In this way, the graphics engine simulates the directional correlation of electromagnetic radiation emanating from surface element 34.

[0038] Back Figure 1Scene simulation 2 further includes lighting 8 and material properties 10. Both involve additional information that complements scene 42 rendered by the graphics engine but is not originally provided by the graphics engine. Lighting 8 includes at least one illuminator 14, i.e., the electromagnetic spectrum of the ambient light illuminating the face element 34. When the face element 34 is illuminated in scene 42 by multiple different types of light sources 46, lighting 8 may include multiple electromagnetic spectra. Each electromagnetic spectrum stored in lighting 8 may be stored as an analytical representation or as a lookup table, wherein storage as a lookup table with sufficiently fine resolution is advantageous.

[0039] Material property 10 is a material database comprising multiple materials, and each material can be configured as a surface element 34 with a given property. A reflectance spectrum S (reference numeral 16) is configured for each of the multiple material properties 10 and stored in the material database. The reflectance spectrum 16 is a material-specific, wavelength-dependent reflectance capability, which gives the spectrum for each wavelength, indicating the proportion of incident light intensity at that wavelength reflected by the material.

[0040] Under the exemplary assumption that face element 34 is a portion of the face of a car driver 44 rendered by a graphics engine, face element 34 can be configured with a material property of "human skin (medium brightness)" according to the desired skin tone of the car driver 44, and due to this configuration, has a reflectance spectrum 16 that mimics the reflectance spectrum of human skin of the corresponding skin tone as measured in reality. The reflectance spectrum 16 configured for material property 10 is only wavelength-dependent. The reflectance spectrum does not take into account the angle dependence of reflection.

[0041] Camera simulation 4 is also not an original component of the graphics engine, but rather a complement to it. The camera simulation should not be confused with the virtual camera 32. The latter only determines the image rendered by the graphics engine, while camera simulation 4 simulates the data channels of a real camera. However, the camera simulated by camera simulation 4 has the same position and spatial orientation in the virtual scene 42 as the virtual camera 32, so the graphics engine's rendering 6 can be used for reprocessing by camera simulation 4. The two can also overlap in process technology, so that the original algorithm of the virtual camera 32 can be directly used in camera simulation 4, or the algorithm of camera simulation 4 can be integrated into the original algorithm of the virtual camera 32. A significant distinguishing feature is that the graphics engine's virtual camera 32 renders a monitor-displayable RGB image of the virtual scene 42 from the perspective of the virtual camera 32, while camera simulation 4 generates camera data from the same perspective based on the virtual scene.

[0042] Camera simulation 4 reads in RGB vectors 12, including color space coordinates, stored by the graphics engine and configured for surface elements 34, and processes these RGB vectors in camera optics simulation 18. Camera optics simulation 18 simulates the process of light incident from a virtual scene onto the camera and imaged onto the sensor pixels of the simulated camera's image sensor. As a result of camera optics simulation 18, for each simulated sensor pixel, there exists an RGB vector 12 for the light exposed to the corresponding sensor pixel. The electromagnetic spectrum is then calculated from each of these RGB vectors 12 using equation (3).

[0043] A scaling factor is applied for each sensor pixel. The calculation routine 20 reads the RGB vector 12 of the corresponding sensor pixel from the camera optics simulation 18. The calculation routine 20 reads one or more illumination lamps of the surface element 34 from the illumination 8, the surface element being imaged onto the corresponding sensor pixel. The calculation routine 20 reads the reflectance spectrum 16 configured for the surface element 34 from the material properties 10, the surface element being imaged onto the corresponding sensor pixel.

[0044] The test specimen 28, in the embodiments described herein, is assumed to be a driver assistance system for monitoring the driver 44 of a vehicle. The driver assistance system is configured to monitor the driver's attention and driving ability using measurable criteria, such as direction of observation, head posture, frequency of yawning or blinking, and to output control data to the periphery of the assistance system 28 when necessary, in order to induce vehicle responses, such as wake-up signals, requests for rest, emergency calls, switching to autonomous driving modes, or emergency braking maneuvers.

[0045] Figure 3 The illustration depicts a virtual scenario 42 set up for testing or training the assistance system. The virtual scenario 42 includes the interior space of a vehicle with a driver 44. The scenario evolves over time: a graphical model of the driver 44 moves and can realistically mimic behaviors relevant to the test subject 28, such as yawning, eyes looking away from the roadway, microsleep, drug or alcohol effects, or medical emergencies such as myocardial infarction, stroke, or shock. A virtual camera 32 is positioned in front of the rearview mirror and simulates an interior space camera aimed at the driver 44 for monitoring the driver 44. The simulated interior space camera detects wavelengths within a certain range. Its electromagnetic spectrum, that is, it operates partly in the near-infrared range.

[0046] The virtual scene 42 also includes infrared lights 46, which provide invisible illumination 14 for the driver 44. The electromagnetic spectrum of the infrared lights serves as the primary illumination in illumination 8. Preservation. Sunlight penetrating the vehicle from the outside serves as a secondary light in Illumination 8. Preserve according to standard daylight (based on the CIE standard colorimetric system).

[0047] Based on material property 10, configure the reflection spectrum for surface element 34. (Human skin, medium brightness). In Figure 1 In the computational routine 20, the unique fundamental function is assigned to the face element 34 according to equation (1) using the information read:

[0048]

[0049] This includes a weighting factor w used to define the relative intensity of infrared illumination compared to sunlight. Basic functions. Describes the angle-independent electromagnetic spectrum reflected by surface element 34. For this fundamental function, computational routine 20 performs the following calculation: the spectral radiance density of the light imaged from surface element 34 onto the image sensor pixel is obtained according to equation (1).

[0050]

[0051] And coefficient According to equation (3)

[0052] The conclusion is as follows.

[0053] All parameters appearing in the integral are known before the start of scene simulation 2, and the number of fundamental functions that computational routine 20 may potentially process is finite, provided that the number of reflectance spectra 16 stored in material property 10 and the number of lighting lamps 14 stored in lighting 8 are finite. Therefore, all integrals that computational routine 20 may potentially calculate, i.e., integrals generated by all possible combinations of lighting lamps and reflectance spectra, can be pre-calculated before the start of scene simulation 2 and stored in memory for retrieval by computational routine 20. The calculation of spectral radiance density 22 is performed via... The determination of these coordinates can then be implemented in real time without any problems. The color space coordinates *r* are stored in an RGB vector I2 and provided by the graphics engine in relation to the angle. Therefore… It is also related to the angle and the final spectral radiance. It is angle-dependent. Therefore, the method can achieve real-time, angle-dependent spectral camera simulation by utilizing the RGB pipeline natively present in the graphics engine. Alternatively, other basic functions can be configured for the surface elements, for example as correction terms, or to take into account the radiation spectrum actively emitted by the surface element 34 when the surface element 34 within the virtual scene 42 is an active radiator.

[0054] Optionally, the calculation can be repeated for the two remaining color space coordinates g and b. This is how the coordinates are obtained for... Three independent values, which may deviate slightly from each other because they are calculated independently, can thus facilitate error minimization. For example, from the obtained values ​​used for Choose one of the following values ​​from the range of values, and for that value, calculate the vector using that value. It has the smallest deviation from the vector provided by the graphics engine.

[0055] The spectral irradiance L (reference numeral 22) is read in by the image sensor simulation 24, which calculates the electrical response of the sensor pixel using the spectral irradiance L calculated for a given sensor pixel and the wavelength-dependent quantum efficiency q of the sensor pixel, and calculates the camera raw data 26 as a whole of the electrical responses of all sensor pixels. In the simplest case, the electrical response of the sensor pixel is derived by the following equation.

[0056]

[0057] This integral also depends only on the wavelength and is therefore similar to the integral in equation (3), which is pre-calculated before the scene simulation begins and stored in memory for 24 reads from the image sensor. K is a variable scaling factor, which is used in particular to take into account the exposure time of the camera being simulated.

[0058] Finally, the synthesized camera raw data 28 is saved using a suitable interface provided for reading through the test piece 28. The test piece 28 processes the camera raw data 28 in the same manner as processing real camera raw data in a real working environment, and outputs control data 30 as a response to the synthesized camera raw data 28. Using the control data 30, a technician can evaluate the characteristics of the test piece 28 by comparing the test piece's response to the synthesized camera raw data 30 with a desired response.

[0059] Optionally, control data 30 (shown by the dashed arrow) may be provided for reading into scene simulation 2, whereby the control data can be taken into account in scene simulation 2. For example, scene simulation 2 may respond to a wake-up signal, including a corresponding animation of the driver 44, requested by means of control data 30, by the scene simulation reopening the driver 44's closed eyes and turning the driver's head back into an upright position.

Claims

1. A computer-implemented method for generating synthetic camera data (26), the method comprising the following method steps: Render (6) the image of the virtual scene (42) and save the color information (12) of at least one face element (34) of the virtual scene (42) in the form of color space coordinates in a multi-stimulus color space. A certain number of basic functions are assigned to the surface element (34), which respectively describe the relative wavelength distribution of the light components emitted by the surface element (34) and model the spectral radiance density (22) of the light emitted by the surface element (34) as a linear combination of the basic functions; The scaling factor of each basic function in the linear combination is derived by calculating the spectral radiance density (22) using at least one color space coordinate from the color information (12) of the surface element and at least one color matching function configured for the color space coordinate; The spectral radiance density (22) is calculated by substituting the scaling factor; and The synthesized camera data (26) is calculated by processing the spectral radiance density (22) in a camera simulation (4) of a camera (32) set in the virtual scene (42).

2. The method according to claim 1, wherein, One of the fundamental functions considers the spectral curve of the illumination (14) of the surface element (34) and the reflectance spectrum (16) of the surface element (34).

3. The method according to claim 2, wherein, Configure material properties (10) for the surface element (34), and configure the reflection spectrum (16) for the material properties (10).

4. The method according to claim 1, wherein the method comprises the following steps: The integral to be solved in order to derive the proportionality factor is calculated in advance; After the pre-calculation of the integral is completed, a scene simulation (2) of the time progression of the virtual scene (42) begins, the scene simulation (2) comprising repeatedly executing the method steps of claim 1. in, The imaging of the virtual scene (42) is an instantaneous state imaging of the virtual scene (42) in each execution. And in each execution, the synthesized camera data is calculated using the pre-calculated integral (26).

5. The method according to any one of the preceding claims, the method comprising the following steps: calculating the color space coordinates (12) taking into account the spatial orientation of the surface element (34) relative to the camera (32), and in particular the angle between the line (40) connecting the camera to the surface element (34) and the surface normal (38) of the surface element (34).

6. The method according to any one of the preceding claims, wherein, The spectral radiance density (22) is calculated in such a way that the spectral radiance density (22) includes wavelengths outside the visible spectrum, especially wavelengths below 380 nm or above 780 nm.

7. The method according to any one of the preceding claims, wherein, The synthesized camera data (26) is calculated as the original synthesized camera data, taking into account the quantum efficiency of the simulated image sensor of the camera (32).

8. The method according to any one of the preceding claims, the method comprising the method steps of: reading in and processing the synthesized camera data (26) by at least one component of an electronic system (28) designed for processing camera data, for training the electronic system (28) or for evaluating the electronic system (28)’s response to the synthesized camera data (26).

9. The method according to claim 8, wherein the method comprises the features of claim 6, wherein, The electronic system (28) is an auxiliary system configured to: monitor the status of personnel (44), especially personnel (44) operating vehicles or machines, analyze camera data from a camera that is pointed at the personnel (44) and capable of detecting wavelengths outside the visible spectrum. The virtual scene (42) includes a virtual representation of the person (44) and the imaging of the virtual scene (42) includes the imaging of the person (44).

10. The method according to claim 9, wherein the method comprises the features of claim 2, wherein, The spectral curve of the illumination mimics the infrared or ultraviolet illumination of the operator (44).

11. A computer program product comprising means for implementing the method according to any one of claims 1 to 10.

12. A computer system, the computer system comprising: Device for calculating synthesized camera data (26) by means of the method according to any one of claims 1 to 10; An interface for providing the synthesized camera data (26) for reading and processing by at least one component of an electronic system (28) designed for processing the camera data (26).

Images