Image acquisition method using an autofocus and associated visualization system or image acquisition device

Abstract

In order to accelerate an autofocus and / or to increase the accuracy of an autofocus, it is proposed to read out two image sensors (2a, 2b), each of which can capture images of an object (18) to be focused, in time alternating during the autofocus, so that respective image data recorded when the respective image sensor (2a, 2b) is read out can be evaluated in order to directly or indirectly determine a best focus plane (7) and to be able to adjust a focusing optics (4) of the associated visualization system (1) accordingly such that sharp images of the object (18), preferably in the form of a respective video image data stream, can subsequently be recorded with both image sensors (2a and 2b).In addition, it is proposed, in addition or alternatively, to operate an image sensor (2a / 2b), which is operated at a certain refresh rate during normal recording of image data, at an increased refresh rate during autofocus in order to accelerate the autofocus.

Description

[0001] The invention relates to an image recording method, wherein image data, in particular in the form of a (respective) video image data stream, is recorded with a first image sensor and a second image sensor of a visualization system (this can be designed, for example, as an endoscopic or microscopic visualization system) which has imaging optics with focusing optics, each (i.e., separately with each of the two image sensors). Before the image data is recorded, an autofocus is performed by detuning the focusing optics. For example, image information can be recorded with the respective image sensor at at least two measuring times, wherein different working or imaging distances are set with an optics group of the visualization system or the said imaging optics at the at least two measuring times.

[0002] The invention further relates to an associated visualization system (in particular its image recording device), which may comprise, in particular, an endoscope or a microscope and a camera control unit (CCU). The visualization system / image recording device comprises a first image sensor, a second image sensor, an imaging optics unit with a tunable focusing optics unit (with which a spatial position of a current focal plane, which is currently being imaged onto the respective image sensor by the imaging system, can be changed), an image processing unit (for processing image data recorded / captured by the respective image sensor), and a controller (for controlling the two image sensors, the focusing optics unit, and the image processing unit).In this system, the controller is provided and configured to record respective image data, in particular in the form of a respective video image data stream and / or in particular in parallel / simultaneously, with each of the two image sensors in an image recording mode and, if necessary, to perform an autofocus by detuning / adjusting the focusing optics before recording the image data in an autofocus mode. The focusing optics can, for example, have axially displaceable focus lenses or lens groups or, for example, at least one tunable lens with a variable focal length. The tunable lens can, for example, be a membrane lens and / or an optofluidic lens or, for example, an elastically deformable elastomer lens.

[0003] It is known from practice that autofocus cyclically captures images with an image sensor while a focusing lens is moved along a travel path (as evenly as possible). The images acquired in this way can then be evaluated for their sharpness (contrast). After the relevant focus area has been optically scanned along the z-axis in this way (= z-scan), the focusing lens can be moved to the adjustment position that provided the highest contrast value, thus capturing the sharpest possible image of the object currently being observed with the system.

[0004] In this approach, the focus group travel speed and the image sensor refresh rate limit the speed at which autofocus can be performed, or the spatial accuracy / resolution, which corresponds to the sampling distance along the travel path of the focusing optics between two captured autofocus images.

[0005] If the refresh rate is set too low, the distances along the focus path may become too large, resulting in the optimal focus point—the current best focus plane—being skipped, meaning it is not located correctly. This can be extremely disruptive for the surgeon, particularly when using a visualization system in medical applications. Conversely, if the refresh rate is increased too much, the maximum shutter time drops significantly, resulting in (i) too little light reaching the image sensor, (ii) no longer achieving sufficiently good image quality during autofocus, and / or (iii) additionally increasing the overhead for data transmission and / or (iv) increasing the overhead for image processing.

[0006] Based on this, the invention is based on the object of enabling an improved use of a visualization system as described above and, for this purpose, of proposing a method and an associated image recording device which can reliably and quickly deliver sharp images even with changing working distances between the image recording device and the object to be recorded.

[0007] To achieve the stated objects, the features of claim 1 are provided according to the invention. In particular, to achieve the stated object in an image recording method, wherein image data, in particular in the form of a video image data stream, is recorded using a first image sensor and a second image sensor of a visualization system which has imaging optics with focusing optics, and wherein an autofocus is carried out by detuning the focusing optics before the image data is recorded, it is proposed that image data from the first and second sensors be evaluated during the autofocus and that, for this purpose, the two image sensors be read out at different times during the autofocus. There is therefore a time offset between the two respective readout times.However, since the focusing optics have moved further during this time interval and thus optically detect a new scanning plane, in such a case the respective focal planes that the respective image sensor sensorily detects / optically scans at the respective readout time differ.

[0008] Alternatively, or in addition to the staggered readout of the two image sensors, it can also be provided that the two image sensors have different imaging distances with respect to the shared imaging optics. For example, the two image sensors can be located at slightly different distances from a shared imaging optics. Due to the different imaging distances, it can be achieved that, for a certain misalignment of the focusing optics, different, spatially spaced focal planes are sensor-detected by the respective image sensor. In other words, for a certain misalignment of the focusing optics, the two image sensors then detect different, spaced focal planes. In such a case, the two image sensors can even be read out simultaneously.But even with simultaneous readout, two different focal planes can be optically scanned / captured with the respective image sensor.

[0009] By means of the autofocus, which can be operated in autofocus mode, the visualization system can thus focus automatically and independently on a new working distance. The method according to the invention offers the technical advantage that the new focal plane on which the visualization system must focus can be found more quickly and precisely than was previously possible. A phase shift between the two image sensors, which arises due to the staggered readout and / or the different imaging distances, can, for example, only / exclusively be active / maintained during autofocus, for example if the two image sensors are read out with a temporal offset only in autofocus mode and not in a subsequent image acquisition mode. The phase shift can, however, also remain during image acquisition mode, in which case the phase shift can then be compensated for in subsequent image processing if necessary.

[0010] As a result of autofocus, the focusing optics can be optimally adjusted to the current focal plane. This setting can then be maintained for both image sensors, ensuring that both image sensors are focused on the current object / focus plane. This allows both image sensors to deliver sharp images as desired in the subsequent image capture mode.

[0011] However, if the two image sensors according to the second alternative have different imaging distances with respect to the imaging optics, so that they capture different focal planes for a given adjustment of the focusing optics, a temporal offset is not absolutely necessary (but possible): Even with simultaneous readout, the two image sensors then scan different focal planes, as mentioned above, which also means that more spatial sampling points along the z-axis (= different z-coordinates of the focal plane just captured) can be obtained per unit of time compared to a situation in which only one of the two sensors is used for autofocus.

[0012] The method according to the invention can also be applied to three or more image sensors that are read at different times and / or arranged at different imaging distances, so that each image sensor captures image data from a different (currently adjusted) focal plane. The more image sensors used in the method, the greater the technical effect achieved in terms of accelerating the autofocus (the time required to find the best focal plane).

[0013] The image acquisition process can therefore switch back and forth between an autofocus mode, in which a z-scan is performed to find a current best focus plane, and an image acquisition mode. As a result of the autofocus, the focusing optics are adjusted so that they focus on the newly determined position z0 of the current best focus plane (which corresponds to a specific adjustment of the focusing optics). In the subsequent image acquisition mode, this adjustment of the focusing optics can then be maintained until a new autofocus is performed. The restart of the autofocus can be triggered by image processing or image analysis (which, for example, detects insufficient image sharpness) that is continuously performed by a controller of the visualization system, or, for example, by an external user input.

[0014] In image acquisition mode, the two image sensors can then preferably be read out synchronously; however, a temporal offset can also be acceptable in image acquisition mode, especially if the respective image data will be merged into a new image using subsequent image processing anyway: For example, in image acquisition mode, a fluorescence image captured with the first image sensor can be superimposed as an "image overlay" on a white light image recorded with the second image sensor. Another example is stereoscopic images, which are each obtained from two individual images captured with the first and second image sensors (preferably at the same time, i.e., synchronously).

[0015] For example, if the two image sensors are operated at a constant and identical frame rate during autofocus, the time offset, or the time interval, between a first image captured with the first image sensor during autofocus and a second, immediately subsequent image captured with the second image sensor will remain constant. However, if the image sensors are operated at different frame rates, this time interval can change during the autofocus process.

[0016] The smaller the acquisition times of the first and second images, the smaller the z-distances that can be resolved with the autofocus. On the other hand, smaller z-distances automatically increase the duration of the z-scan for a given length in the z-direction that is to be scanned optically and stepwise with the autofocus.A possible embodiment of the method therefore proposes that the position z0 of the current best focus plane is initially roughly determined at a fast focus group speed and / or at a first (slow) frame rate using both image sensors (first z-scan, performed with both image sensors). Subsequently, for a more precise determination of z0 (position of the best focus plane), a second z-scan, for example, a fine scan, is performed again with both image sensors at a slow focus group speed and / or at a second frame rate that is higher than the first frame rate. This allows for both fast and accurate location.

[0017] In a further advantageous embodiment, it can be provided that during autofocus, by adjusting the focusing optics, a position of a focal plane of the imaging optics is shifted in a z-scan along an optical z-axis in order to set / find a current best focus plane to be used in the subsequent acquisition of the image data. Alternatively or additionally, it can be provided that the autofocus is executed to determine a position z0 of a / the current best focus plane. This allows reliably sharp images to be recorded in the subsequent image acquisition mode.

[0018] Alternatively or additionally, autofocus can be performed to determine an adjustment of the focusing optics that corresponds to a / the current best focus plane. A significant advantage of this is that the reaction time required by the system to automatically respond to changing working distances by adjusting the focus position can be reduced. This significantly increases the ease of use of the visualization system.

[0019] In a further advantageous embodiment, it can be provided that during the autofocus a first image is recorded with the first image sensor at a time t1 and a second image is recorded with the second image sensor at a time t2, in particular while the focusing optics are being adjusted, preferably in such a way that a respective position z1 and z2 of the focal plane differs at the times t1 and t2.

[0020] The advantage of this type of autofocus design is that the effective frame rate available for quickly and accurately locating the best focus plane is increased compared to the frame rate at which the respective image sensor is operating. This is because the temporal offset of the respective recording times at which images are captured with the first and second image sensors during autofocusing results in an overall image sequence available for evaluation in which the time intervals between the individual images are shortened. This increases the resolution in the z-direction that can be achieved when determining the position z0 of the current best focus plane.

[0021] To determine the current best focus plane, a parameter relevant to the autofocus, for example, image contrast and / or image sharpness of the image sequence captured with the respective image sensor, can be evaluated in a conventional manner. However, the following autofocus methods can also be used by the visualization system to determine the best focus plane: a) contrast autofocus (a contrast value is used to determine the plane with the highest sharpness); b) autofocus via phase detection (by comparing the phase difference between two images). Such approaches can be implemented at the sensor level of each of the image sensors used by the system, provided the respective image sensor supports such a function, or by image analysis of the two images of a stereo system or, for example, by analysis of two images that are separately captured by the visualization system in different spectral ranges.

[0022] In particular, it can be provided that during autofocus, a sequence of images is captured with each of the first and second image sensors, which are then evaluated together to determine the best focus plane. Images of the first and second sequences differ in terms of their respective acquisition times and / or their respective corresponding focus plane (this corresponds to the z-coordinate on which the focusing optics focuses at the time of acquisition).

[0023] In a further advantageous embodiment, it can be provided that a time offset t2-t1 between successive measurement times t1 and t2, at which the first image and the second image are respectively recorded, is at least 10%, preferably more than 25%, of a time interval between two directly successive individual images of the first or second image sequence.

[0024] For example, if the frame rate of the first image sequence is constant at 50 Hz, a single image is captured with the first image sensor every 20 ms. The temporal offset should then be at least 2 ms (= 10%), but preferably at least 5 ms (= 25%). The most uniform spatial sampling of the focal planes along the z-axis is achieved with a z-scan performed at almost continuous speed when the temporal offset is approximately or exactly half the temporal interval between two images, when both image sensors are operated at the same frame rate. This is because the scanned focal planes then exhibit approximately equal spatial distances along the z-axis.

[0025] Preferably, the images of the respective sequence can be captured at a constant frame rate. This frame rate can correspond to the maximum frame rate possible with the respective image sensor at the currently set image resolution.

[0026] In a further advantageous embodiment, it can be provided that, within the scope of the autofocus, a first image sequence is recorded with the first image sensor and a second image sequence is recorded with the second image sensor. To determine the position z0 of the current best focus plane, both image sequences are preferably evaluated together, in particular by comparing individual images of the two image sequences.

[0027] Particularly advantageously, the corresponding best focus plane of the recorded image sequences can be quickly detected, for example by comparing the individual images of the two image sequences.

[0028] In a further advantageous embodiment, it can be provided that the recording of the first image sequence is triggered in time with a first synchronization signal and the recording of the second image sequence is triggered in time with a second synchronization signal, and wherein the first and the second synchronization signals show a time offset.

[0029] It is particularly advantageous, for example, to be able to evaluate the image information at different measurement times.

[0030] Alternatively or additionally, it may be provided that the second synchronization signal is generated by means of a phase shift of a jointly used synchronization signal.

[0031] For example, it can be provided that the phase shift is activated during autofocus or permanently used by the sensors and compensated for, for example, in subsequent video processing. This can be particularly advantageous, for example, to ensure that the entire sensor arrangement with at least two sensors behaves like a single sensor with a higher frame rate.

[0032] For example, it can further be provided that the first image sensor is triggered with the common synchronization signal (which then serves as the first synchronization signal), while in a second signal path, the second synchronization signal is generated from the common synchronization signal by means of the phase shift. A phase shift can be understood here, in particular, as meaning that the synchronization signal is delayed in time. A phase shifter device can be configured for this purpose.

[0033] In a further advantageous embodiment, it can be provided that image data recorded during the autofocus with the first image sensor and / or with the second image sensor are first adapted via an image processing algorithm before the image data thus adapted are evaluated within the scope of the autofocus, in particular wherein the image data are adapted with regard to an optical perspective and / or processed in such a way that a comparison of the first and the second image is simplified.

[0034] For example, if the visualization system is designed for stereoscopic vision, so that the first and second image sensors record an object to be observed from (slightly) different spatial directions, the images of the respective image sequence may differ (slightly) with regard to the scene currently being observed with the visualization system, for example with regard to image detail or perspective / angle of view. Such deviations can be taken into account when evaluating the two image sequences. For example, each of the two image sequences can be subjected to a respective image processing algorithm that makes the images more comparable with each other, for example by only comparing respective partial sections of the recorded images with regard to an evaluation criterion (e.g. "optimal image sharpness").This can, for example, simplify a contrast comparison of the images, which can be advantageous for quickly and reliably finding the best focus plane.

[0035] In this method, for example, a third image sequence can be generated from the first image sequence and the second image sequence, in which respective images from the first and second image sequences follow one another. The third image sequence can then be evaluated accordingly to determine the position z0.

[0036] During autofocus, the first sensor can have or be operated at a refresh rate that matches the refresh rate of the second sensor. However, different refresh rates can also be used, in which case the time offset between the first images of the first image sensor and the second images of the second image sensor varies.

[0037] In a further advantageous embodiment, it can be provided that an adjustment speed of the focusing optics is changed during the autofocus, preferably wherein the adjustment speed of the focusing optics is slowed down when passing through the z-scan when approaching the current best focus plane.

[0038] If the frame rate of both image sensors is kept constant in such a procedure, a higher spatial resolution can still be achieved in the area of ​​the best focus plane when determining the position z0, because with a lower adjustment speed and the associated (momentary) lower speed of the z-scan, the spatial distance between the individual focus planes optically scanned with the two image sensors decreases, so that the possible spatial resolution in the z-direction increases.

[0039] Furthermore, it can be provided, for example, that during a movement and / or adjustment of the focusing optics, intermediate images are generated by the second or further sensors in relation to the first / the other sensors.

[0040] It is particularly advantageous, for example, by increasing the adjustment speed of the optical group, to achieve a sharp image more quickly and / or to evaluate the recorded image information more quickly with regard to the highest contrast and to reach an autofocus point more quickly.

[0041] In a further advantageous embodiment, it can be provided that the method is carried out fully automatically by a controller of the visualization system, in particular without user input, in particular as soon as the controller independently detects that a current image quality of the image data just recorded with the visualization system no longer meets specified requirements.

[0042] In other words, the controller can be configured to detect, based on image analysis of the image data recorded in image acquisition mode, when the working distance between the visualization system and an object currently being observed has changed so significantly that refocusing using autofocus makes sense. In this case, the controller then initiates autofocus mode and performs autofocus. Such initiation of refocusing can also be triggered, for example, by a sensor that detects a change in the working distance, and / or by an external system, such as an external camera system, and / or by a robotic arm movement (where the robot arm moves the two image sensors and thus changes the working distance).Accordingly, in such a case, the controller can interrupt the imaging (but it does not necessarily have to interrupt it) and can then execute the autofocus so that image data with sufficient image sharpness can be recorded again.

[0043] In a further advantageous embodiment, the first image sensor and the second image sensor can be implemented as two non-overlapping, preferably adjacent, regions of an active sensor surface of a single image sensor chip of the visualization system. In such a case, the two sensor surfaces are thus located in one plane and thus typically at the same imaging distance from the imaging optics. Therefore, the two regions of the single image sensor chip are read out at different times during autofocus, thus achieving the advantages of the invention.

[0044] Such a design of the method is suitable, for example, for a stereoscopic visualization system in which only a single image sensor (with a correspondingly large format) senses two parallel optical channels / two imaging beam paths used for stereoscopic imaging. In such a case, with a correspondingly large dimensioning of the image sensor, one half of the active sensor surface can sense a left optical channel and an adjacent second half of the active sensor surface can sense a right optical channel of the visualization system. Even in such an application, the temporally offset readout according to the invention can be used to increase the accuracy and / or speed of the autofocus.In this case, during autofocus (as well as during the subsequent actual imaging), individual images are recorded with the respective area of ​​the single image sensor, but at different times during autofocus, in particular while the focusing optics are adjusted.

[0045] In one embodiment, it can therefore be provided that the two image sensors are used to record stereoscopic images.

[0046] A particular advantage is that the time-shifted readout of the image sensors only occurs during autofocus mode (improved approximation to the best focus plane), while in stereoscopic vision, simultaneous readout of both image sensors can be used to simultaneously acquire image pairs for stereo imaging.

[0047] Alternatively or additionally, the two image sensors can be used to sensor-capture spectrally different wavelength ranges. In this case, in particular, an imaging beam path of the imaging optics can be split by means of a spectrally selective beam splitter into a first imaging beam path, captured by the first image sensor, and a second imaging beam path, captured spatially separately by the second image sensor.

[0048] It should be expressly mentioned here that both approaches can be combined, so that even with stereoscopic imaging, for example, a spectral image and a white light image can be recorded simultaneously using a visualization system configured according to the invention. If such a system has a total of four image sensors (e.g., two monochrome image sensors for recording respective spectral images and two color image sensors for recording respective white light images), all four image sensors can be read out with a temporal offset during autofocus such that at a given recording time, only one of the four image sensors records an image. This ideally improves the spatial resolution of the autofocus by a factor of four compared to using only one sensor at its maximum frame rate.

[0049] Alternatively or additionally, the features of the independent claim 12 directed to an image recording method are provided according to the invention to achieve the stated object. This method provides that image data (in particular in the form of a video image data stream) are recorded in an image recording mode using at least one image sensor of a visualization system having imaging optics with focusing optics, and that, before recording the image data in the image recording mode, an autofocus is carried out by detuning the focusing optics.To achieve the stated object, the invention proposes that during autofocus, the at least one image sensor is read out at different measurement times, and that at each of these measurement times, less image data (particularly due to a reduced color depth compared to the image acquisition mode) is read out and used for evaluation than is read out for individual images in the subsequent image acquisition mode. However, this basic approach can also be used in the previously explained method according to claim 1, i.e., even when using two or more image sensors during autofocus.

[0050] Alternatively or additionally, the invention can provide for the at least one image sensor to be operated during autofocus at a first refresh rate that is higher than a second refresh rate at which the at least one image sensor is operated in the subsequent image capture mode. This also allows the amount of image data to be reduced, which can then be used to accelerate the autofocus because this reduced image data can be read and processed more quickly. This is particularly suitable for systems with limited computing power.

[0051] With this approach according to claim 12, the frame rate (more precisely, the temporal frequency at which the respective image sensor captures respective image data at different focal planes) can be increased by reducing the spatial resolution of the image sensor used during autofocus. However, if an image sensor with the same resolution already supports higher frame rates than those required for subsequent imaging (typically 60 Hz), this can be exploited during autofocus. In image capture mode, however, multiple frames can be removed if necessary to generate a desired frame rate on the monitor / in the video image data stream.

[0052] By utilizing a higher refresh rate than the image capture mode, the image sensor can also be switched to a lower (spatial) resolution during autofocus, thus enabling an even higher refresh rate. It can also be advantageous to operate the image sensor in a so-called "binning mode," in which neighboring image elements or image regions, or individual pixels, are combined. This improves the signal-to-noise ratio, with pixels being able to be combined row-by-row, column-by-column, or block-by-block. Such approaches can at least partially compensate for the reduced sensitivity of the image sensor that automatically accompanies a higher refresh rate. Furthermore, the positive effect on the speed and spatial accuracy of the autofocus described above can be further enhanced.Finally, all of this can be applied to each of the at least two image sensors if the system has at least two image sensors, which can then be read out at different times. However, increasing the frame rate during autofocus can also be used on its own with just a single image sensor to speed up autofocus.

[0053] In this method according to claim 12, it can also be further provided that, by adjusting the focusing optics, a position of a focal plane of the imaging optics is shifted in a z-scan along an optical z-axis in order to set a current best focus plane. Likewise, the autofocus can also be used here to determine a position z0 of the current best focus plane and / or an adjustment of the focusing optics that corresponds to the current best focus plane.

[0054] Autofocus can be terminated once the z0 position has been determined and / or the focusing optics have been adjusted to the currently found best focus plane (the actual calculation of the z0 value is not mandatory). With the resulting focusing optics adjustment, high-quality image data can then be acquired in image acquisition mode.

[0055] A particular advantage of this method is that less data needs to be read out at each measurement time during autofocus (= autofocus mode) than during the actual image acquisition. This makes it possible to operate at least one image sensor at a higher frame rate during autofocus than during the actual image acquisition (image acquisition mode), thus achieving a higher spatial z-resolution.

[0056] If this method uses two image sensors which, in particular as described above, can be read out at different times during autofocus, each of these image sensors can be operated at a higher refresh rate than during the subsequent recording of image data in image acquisition mode. In other words, during autofocus, each of these two image sensors can be operated at a first refresh rate that is higher than a respective second refresh rate at which the respective image sensor is operated in the subsequent image acquisition mode. And of course, this approach is also applicable to a visualization system that has only a single image sensor, but can record at least two different images, for example for stereoscopic vision, with the single image sensor simultaneously (e.g. due to a division of the active sensor area into at least two image areas).

[0057] As already mentioned, to determine position z0 of the current best focus plane, an autofocus value, such as a contrast and / or sharpness of the respective image information recorded in the autofocus, can be read out and evaluated.

[0058] In a further advantageous embodiment, it can be provided that a respective image information which is captured with the at least one image sensor at the different measuring times is a reduced image information which is obtained by reading out and not reading out pixels of the at least one image sensor line by line and / or column by column, in particular alternatingly.

[0059] With this method, only a single image sensor can be used for autofocus under certain circumstances, yet an improvement in autofocus can still be achieved. This is because the image sensor in question, which is read out with less image information and / or a higher frame rate in autofocus mode, can thus be one of at least two sensors in the visualization system used (e.g., a color image sensor, while the visualization system also has another monochrome image sensor that is operated in parallel with the color image sensor during the actual image acquisition, for example, to capture a spectral image, but is not used for autofocus, for example).

[0060] A reduction in the amount of image information to be processed in autofocus mode can be achieved in a variety of ways. For example, the image information determined at the respective measurement time with the at least one image sensor during autofocus can be based on a subset of image pixels of the image sensor. Such a subset can be defined, in particular, by a subset of rows and / or columns of the image sensor and / or by a specific segmentation of the active sensor area of ​​the image sensor.

[0061] The generally available image information, which defines a maximum image resolution, can be predetermined by a maximum number of image pixels in the sensor array. In such a case, reduced image information can be created, for example, by combining several image pixels into a larger unit. This can be achieved, for example, by so-called "pixel binning," in which, for example, neighboring pixels are combined, in particular with the help of software, to create a new (virtual) pixel that results in particular from the sum of the individual combined pixels. Furthermore, several image pixels can be combined into a larger unit, for example by "cropping" and / or "windowing / framing."All such configurations can lead to a reduction in the image information to be processed in autofocus mode, thus achieving a higher frame rate and thus a higher spatial resolution in the optical scanning and / or a faster autofocus (for a given adjustment speed of the focusing optics).

[0062] The inventive approach therefore proposes, in particular, that the respective image information acquired at the respective measurement time to determine the current best focus plane be reduced compared to the image information acquired in the subsequent image acquisition mode with the same sensor. For example, the resolution of the images acquired sensorically with the visualization system during autofocus, more precisely with the at least one image sensor, can be reduced compared to the image resolution used in the subsequent image acquisition mode.

[0063] The visualization system can, for example, have a switching device that enables the implementation of the autofocus function. If the visualization system has two image sensors, for example, the switching device can be configured to switch at least one of the two image sensors during autofocus to an autofocus mode in which only a portion of the image information that is generally available to the sensor is read out at a specific measurement time during autofocus. In this case, the switching device can preferably also increase the frame rate of the affected image sensor, so that images are recorded at shorter intervals with the affected image sensor during autofocus than in the subsequent image acquisition mode.

[0064] Regarding the visualization system, it should be noted that it can of course be equipped with a focus actuator for adjusting the focusing optics and / or a zoom actuator for setting an optical zoom.

[0065] Alternatively or additionally, to achieve the stated object, the features of the independent claim directed to a visualization system are provided according to the invention. This visualization system can comprise an endoscope or a microscope; it further comprises a first image sensor, a second image sensor, imaging optics with tunable focusing optics, an image processing unit, and a controller. The controller can be configured to record respective image data, in particular in the form of a respective video image data stream, in particular in parallel / simultaneously, with each of the two image sensors in an image recording mode and, if necessary, to perform an autofocus by detuning the focusing optics in an autofocus mode before recording the image data. For example, it can also be provided that the controller outputs / forwards recorded image data to an external device.

[0066] To achieve the stated object, it is further proposed according to the invention that the controller is configured to carry out a method as described above and / or claimed here according to one of the claims.

[0067] This visualization system makes it extremely advantageous to realize the aforementioned advantages of the image acquisition method according to the invention. This makes it possible, for example, to implement the aforementioned advantages in the context of an endoscopic or stereoscopic procedure, making the entire procedure easier and more efficient because the autofocus is performed reliably and quickly by the system, automatically.

[0068] It has already been mentioned that the two image sensors can be arranged spatially separated from each other or can be formed by a respective region of an active sensor surface of a single image sensor chip. Alternatively or additionally, the two image sensors can be provided to sensor-detect a respective imaging beam path generated by the shared imaging optics.

[0069] It is particularly advantageous to use different image sensors, for example. For example, it can also be provided that the two image sensors are connected to a common data stream channel. This makes it possible to advantageously detect and visualize rapid movements and / or changes in the recorded image.

[0070] Alternatively or additionally, it can be provided that the two image sensors sensorily detect a spectrally different imaging light, in particular after splitting by means of a spectrally selective beam splitter.

[0071] The two image sensors can, for example, also be arranged in relation to a respective imaging optics so that both image sensors capture an identical object plane. The advantage of such a configuration is that the two imaging optics can create two separate optical channels, which can be used for stereoscopic imaging (cf. Fig. 14).

[0072] Alternatively or additionally, the two image sensors can offer different spectral sensitivities. This is advantageous because different wavelength ranges can be captured / imaged by the image sensors and / or the quality of the image data recorded by the image sensors can be increased, for example, by adapting the spectral sensitivities to each other, and the wavelength range that can be imaged by the image sensors can be expanded.

[0073] Alternatively or additionally, it can be provided that the two image sensors offer different spatial resolutions and / or that the two image sensors offer different refresh rates.

[0074] The invention will now be described in more detail using exemplary embodiments, but is not limited to these embodiments. Further developments of the invention can be derived from the following description of a preferred embodiment in conjunction with the general description, the claims, and the drawings.

[0075] In the following description of various preferred embodiments of the invention, elements which correspond in function are given the same reference numbers even if they have a different design or shape.

[0076] It shows: Fig. 1 an autofocus method for finding a best focus plane, as is previously known from the prior art, Fig. 2 shows the internal structure of an imaging and focusing optics as can be used in a method according to the invention, Fig. 3 a first example of how an image recording method according to the invention can be advantageously used in a visualization system with two separate image sensors, Fig. 4 shows another example of a method according to the invention, Fig. 5 illustrates the main technical advantages of the AF approach according to the invention, Fig. 6 shows different operating modes in which the respective image sensor of a visualization system according to the invention can be operated in different modes, Fig. 7-9 illustrate different ways of reading an image sensor of a visualization system according to the invention, Fig. 10 shows a typical application situation in which a visualization system according to the invention can be used effectively, and finally the Fig. 11-16 show possible embodiments of visualization systems according to the invention, each of which can be configured to carry out an autofocus method according to the invention.

[0077] Fig. 1 schematically shows the situation in which an object 18, for example a specific surgical area, is observed with a visualization system 1 such as a microscope, wherein the object 18 is currently located at a specific working distance 17 measured from the first optical surface of the system 1. The system 1 clearly has an imaging optics 3, each of which generates imaging beams that can be recorded / sensor-captured using two different image sensors. In order to record sharp images of the object 18, the system 1 performs an autofocus in a manner known per se, wherein a focus lens is detuned for this purpose such that the current focal plane 5 (i.e. the plane in object space that is sharply imaged onto the respective image sensor 2 for the given current position of the focus lens) shifts along the optical z-axis 6 shown.At the same time, the respective image contrast of the recorded image is recorded electronically, so that the image in the lower half of the . Fig. 1 results. It can be seen that for object 18, a so-called best focus plane 7 exists at the z-coordinate z0, which provides the best image contrast. The points on the curve shown illustrate the different focus planes 5 that were optically scanned with system 1. During such an optical scan to determine the best focus plane, only one image sensor is read.

[0078] Fig. Figure 2 illustrates how such a z-scan can be optically implemented. As can be seen, during autofocus, a focusing optics 4 as part of a more complex imaging optics 3 (which generates images of the object 18 on the image sensor 2 shown) is detuned using a focus actuator 22, which in turn is controlled by a controller 15. The system 1 also has a zoom actuator 21, which can be used to adjust an optical zoom during imaging. The image contrast values ​​to be determined during autofocus (AF) are determined by an image processing unit 14, which electronically reads the associated image sensor 2.

[0079] Performing such a Z-scan is complex and therefore takes a certain amount of time. In a specific application situation, such as in Fig. 10 when the visualization system 1 is used to observe an operation area and the working distance 17 changes very frequently (as soon as the Fig. 10, which carries the system 1, is moved), it would be of great advantage to accelerate the autofocus and thus switch back to the actual image recording mode as quickly as possible, in which high-resolution and sharp images of the surgical area can then be recorded as a live video stream with the help of the system 1.

[0080] Fig. 3 shows a first example of an image recording method designed according to the invention or of a visualization system 1 designed according to the invention. This system 1 already has, like that of Fig. 1 has two separate image sensors 2a and 2b and an associated imaging optics 3 with a focusing optics 4. With each of the two image sensors 2a and 2b, a respective video image data stream can be generated, i.e. a sequence of images, whereby the individual images can be captured at different times with the respective sensor 2a / 2b.

[0081] Such a system 1 could, for example, be as in Fig. 11, with a proximal first image sensor 2a and a second distal image sensor 2b. Starting from a cover glass 30, which enables an oblique view, the imaging beams first pass through a complex imaging optics 3 consisting of several rod lenses 32 and then enter a spectrally selective beam splitter 10, which has a spectrally selective mirror surface 31 (cf. Fig. 11). This allows spectrally different images to be recorded with the two image sensors 2a and 2b. For example, the beam splitter 10 can be configured so that the image sensor 2a records white light images, i.e., visible wavelengths, while the second image sensor 2b captures invisible spectral images in the near infrared.

[0082] One can also see in Fig. 11, that with the aid of the beam splitter 10, the imaging beam path generated by the imaging optics 3 is split into a first imaging beam path 11a, which is captured by the first image sensor 2a, and a second imaging beam path 11b, which is captured by the second image sensor 2b. This allows the two image sensors 2a and 2b to each capture spectrally different wavelength ranges.

[0083] The Fig. 12 shows the visualization system 1 of the Fig. 11 in a bottom view. As can be seen, the visualization system 1 has two separate optical channels 34a and 34b, so that each of the two spatially separated image sensors 2a, 2b can capture stereo images in the respective spectral range.

[0084] The Fig. Figure 13 shows the spectrally selective beam splitter 10 with the internal mirror surface 31 in detail, which is formed by a first prism 33a. The imaging beams generated by the imaging optics 3 enter the first prism 33a through the entrance surface A, with visible wavelengths passing through the mirror surface 31, then passing through the second prism 33b, and finally exiting from the end surface B. Wavelengths in the NIR range, however, are reflected by the mirror surface 31 and, after further internal reflection (cf. Fig. 11), from the exit surface C of the front prism 33a.

[0085] As can be seen from the dashed lines in the Fig. 3, in the method according to the invention, when executing the autofocus, which takes place before the actual images are taken in the later image recording mode, not only image data from the first image sensor 2a but also additionally image data from the second image sensor 2b are evaluated. In this case, the two image sensors 2a and 2b are read out alternately and at different times, while the focusing optics 4 are detuned, ie while the z-scan is being carried out. Therefore, the two image sensors 2a and 2b alternately optically scan different focal planes 5 at different focal distances. If the respective step size (compare the block arrow in Fig. 3) chosen to be similarly small as in the example of Fig. 1, the number of z-positions that can be optically scanned during the Z-scan is approximately doubled because two image sensors 2a, 2b are now used in parallel.

[0086] As a result, the effective step size (black block arrow) that can be achieved using the method according to the invention is reduced. This allows for a more precise resolution in the z-direction when determining the best focus plane 7. In the lower graph of the Fig. 3, the black points belong to the first image sensor 2a, while the optical sampling points, which are shown as circles in the graphs, belong to the second image sensor 2b.

[0087] Unlike the state-of-the-art example according to Fig. 1 is used in the example of Fig. 3 For example, to capture the focal plane 5 at position z1, the first image sensor 2a is read out at a time t1, while the subsequent focal plane 5 at position z2 is optically scanned by the second image sensor 2b being read out at a different time t2, after the focusing optics 4 have been adjusted accordingly. Accordingly, not only a spatial distance Δz results between the two focal planes 5, but also a temporal offset Δt = t2-t1 between the first image, which is recorded with the first image sensor 2a while the focusing optics 4 is focusing on position z1, and the second image, recorded with the second image sensor 2b while the focusing optics 4 is focusing on position z2.

[0088] Not shown in the figures is a further embodiment of the method according to the invention in which both of the Fig. 3 are read out simultaneously, but (e.g. in the z-direction shown) show a small spatial offset Δz (i.e. are arranged spatially offset from one another), so that the image sensors 2a, 2b are arranged at different imaging distances from the imaging optics 3 and therefore, during simultaneous readout, still optically scan spaced focal planes 5. In this case, at each scanning / readout time, two different focal planes 5 are always spatially recorded with the two image sensors 2a, 2b, the spatial distance of which results from the optical imaging equation taking into account the offset Δz. As a result, an equally high spatial resolution in the determination of z0 can be achieved in this way, similar to the example of the Fig. 3.

[0089] As a comparison of the Fig. 1 and Fig. As shown in Figure 4, the step size of the respective z-scan, which is realized using the respective image sensor 2a / 2b, could also be increased. In this case, despite the double step size (compared to Fig. 1) the same number of z-positions are optically scanned, because now two image sensors 2a and 2b are used in parallel but phase-shifted to each other and / or spatially offset to each other during the optical scanning of the focal planes 5. If in such a case, for example, the frame rate is increased and the travel speed of the focus lens 4 is increased, the Fig. 4 can be performed in a shorter time. This time saving (compared to the approach used in the state of the art according to Fig. 1, where only a single sensor is used during autofocus or two image sensors that are read out simultaneously) is in the Fig. 5 illustrates: Fig. Figure 5 illustrates the time required to perform a complete z-scan and determine the best focal plane 7. The dashed line shows the corresponding curve for the inventive approach.

[0090] Fig. Figure 6 illustrates the two different modes in which a visualization system according to the invention can be operated. In normal image acquisition mode, the respective image sensor 2 is exposed for a specific exposure time A and then electronically read out within a specific readout time B. The total time period A+B thus corresponds to the time required to capture one image (frame). According to the inventive approach, it can now be provided that when the system 1 switches to AF mode, the image sensor 2, which is used for optically scanning the focal planes 5, is operated at an increased frame rate, as shown in the upper half of the Fig. 6. In this case, the image sensor 2 can, for example, due to lower resolution, record two images within a time period that is normally used to record one image of a continuous video image data stream. It is understood that this also makes it possible to achieve a higher z-resolution or an acceleration of the autofocus even when using only a single image sensor, because the higher frame rate either allows more optical scanning points to be acquired along the z-axis 6 or allows a certain number of optical scanning points to be traversed more quickly along a predetermined distance along the z-axis 6.

[0091] The Fig. 7-9 illustrate how such a higher frame rate can be achieved during autofocus with a conventional image sensor 2. In normal image capture mode, the image sensor 2 can, for example, be used as in Fig. 7, so that all pixels 23 of the image sensor 2 are read out. However, using pixel binning, several pixels 23 can be combined into pixel areas 24 in autofocus mode. While this reduces the spatial resolution of the image sensor 2, it also allows the image sensor 2 to be electronically read out more quickly, allowing a higher frame rate to be achieved during autofocus.

[0092] In addition or as an alternative to this, as in Fig. 9, only a specific area 13 of the image sensor 2 (then, for example, with full resolution) can be read out during autofocus. In the example of the Fig. 9, for example, outer image areas were ignored (image cropping). This also allows a higher frame rate of the image sensor to be achieved. Another option, which can be used in addition or as an alternative, is to reduce the color depth with which the image sensor electronically captures images while autofocus is being performed. This can also speed up the time it takes to capture images during autofocus. During the actual subsequent imaging, after autofocus has been performed, the color depth can then be increased again, with images then being recorded at a lower frame rate.

[0093] As in the Fig. 3 and Fig. 4, in order to determine the best focus plane 7 or for the best possible adjustment of the focusing optics 4 (in order to produce a sharp image of the object 8), both image sequences 8a and 8b, which are supplied by the first and second image sensors 2a / 2b, respectively, are evaluated together by an image processing unit 14 of the system 1.

[0094] In the Fig. 3 and Fig. In the examples shown in Figure 4, the two image sensors 2a, 2b are operated with a phase shift of 180°, which achieves spatially uniform scanning because the focal planes 5 exhibit the same spatial offset on both sides. Such an approach is recommended when the position of the best focal plane is not precisely known.

[0095] Fig. 16 illustrates that even if only a single image sensor chip 16 is present, which is read out during autofocus, an acceleration can already be achieved if the image sensor 2 (e.g. at lower resolution) is read out at an increased frame rate.

[0096] The inventive visualization system according to the Fig. 14 (side view) and 15 (bottom view) has, like that of the Fig. 16 only has a single image sensor chip 16. However, when autofocusing, the Fig. 14 and Fig. The left region 13a of the active sensor surface 12 of the image sensor chip 16, illustrated in Figure 15, is used as a first image sensor 2a, and the second region 13b as a second image sensor 2b. If these two adjacent but non-overlapping regions 13a, 13b are read alternately and with a time delay while the focusing optics 4 are adjusted, images can be recorded with the two regions 13a, 13b that optically scan respective different focal planes 5.

[0097] It should be noted that this visualization system 1 also complies with the Fig. 14 and Fig. 15 has two optical channels 34a and 34b, so that each of the two image areas 13a and 13b can sensorily capture a respective individual image of the observed object 18 (generated using the respective optical channel 34a / 34b). These respective individual images differ slightly in perspective with respect to the object 18, which can be exploited in the image recording mode, as is known per se, to record stereo images. During autofocus, however, these different perspectives can be taken into account by the image processing unit 14, so that even based on such slightly differing images, the focal plane 5 for which the best image sharpness can be achieved can be identified.

[0098] In previously known autofocus methods, only one sensor is used for autofocus in a stereo system, which is why such an offset of the two images is not a problem. In a stereo system such as the one in the Fig. 14 can, however, be used to ensure that the two image areas 13a and 13b become perspectively congruent as soon as the focal plane is reached.

[0099] It would be conceivable, in each of the two stereo images (taken with image areas 13a and 13b respectively - cf. Fig. 14) to overlay a circular area or other object (using image processing). These objects / circular areas will then merge as soon as the two optical channels are in focus. This reflects the fact that, depending on the system configuration, the focal plane is also the convergence plane (both optics show (at least approximately) the same image section). However, these two planes can be set differently optically or through subsequent image processing.

[0100] Some previously known systems, which still use an eyepiece instead of a screen to visualize the observed object area, use a laser that is coupled into the two optical channels. In the focus and thus also in the convergence plane (usually the planes are adjusted so that they coincide), two laser points are then congruent, and the user receives direct visual feedback that the focal plane lies where the two laser beams overlap. According to the invention, however, this previously known approach can also be created artificially without the use of a physically present laser by simply superimposing an object in the image using signal processing. This is because the superimposed objects then also overlap in the focal plane.The invention therefore proposes, in particular, to display a virtual object in both image areas in order to provide direct visual feedback to the user, which allows a conclusion to be drawn about the position of the current focal plane.

[0101] It should also be mentioned that the Fig. 14 and Fig. 15, the optics shown are divided into an image area 13a on the left and an image area 13b on the right (cf. Fig. 14). However, it is quite conceivable and possible to arrange the two optics / the two optical channels not next to each other (i.e. along the y-axis in Fig. 14), but one above the other (along the x-axis in Fig. 14). Even in such a case, only a single image sensor can be used, which then has a first active region 13a and a second active region 14a arranged underneath, in order to realize a visualization system 1 according to the invention.

[0102] In summary, in order to accelerate an autofocus and / or to increase the accuracy of an autofocus, it is proposed to read out two image sensors 2a and 2b, each of which can capture images of an object 18 to be focused, in time alternating during the autofocus, so that respective image data recorded when the respective image sensor 2a, 2b is read out can be evaluated in order to directly or indirectly determine a best focus plane 7 and to be able to adjust a focusing optics 4 of the associated visualization system 1 accordingly so that subsequently sharp images of the object 18 can be recorded with both image sensors 2a and 2b, preferably in the form of a respective video image data stream.In addition, or alternatively, it is proposed to operate an image sensor 2a / 2b, which is operated with a certain refresh rate during normal recording of image data, with an increased refresh rate / frame rate during autofocus in order to accelerate the autofocus (cf. Fig. 3). List of reference symbols 1 visualization system (particularly designed as an endoscope or microscope or macroscope) 2 image sensors 3 Imaging optics 4 Focusing optics 5 Focal plane 6 z-axis 7 Best focus level 8 image sequence 9 Synchronization signal (to trigger image capture) 10 spectrally selective beam splitters 11 Imaging beam path 12 active sensor area (of 2 / 16) 13 areas (of 12) 14 Image processing unit 15 controllers 16 image sensor chip 17 Working distance 18 objects 19 Contrast value 20 focus distance 21 Zoom actuator 22 Focus actuator 23 pixels 24 pixel areas 25 operating table 26 patients 27 Robot arm 28 screen 29 Microscope 30 cover glass 31 spectrally selective mirror surfaces (out of 10) 32 rod lens 33 Prism 34 optical channels z0 spatial position of 7

Claims

[1] Image recording method, wherein image data, in particular in the form of a video image data stream, are recorded with a first image sensor (2a) and with a second image sensor (2b) of a visualization system (1) which has an imaging optics (3) with a focusing optics (4), and - wherein an autofocus is carried out by detuning the focusing optics (4) before the image data is recorded, characterized by , - that image data from the first and second image sensors (2a, 2b) are evaluated during the autofocus and - that the two image sensors (2a, 2b) are read out at different times during the autofocus and / or - wherein the two image sensors (2a, 2b) have different imaging distances with respect to the common imaging optics (3), so that the two image sensors (2a, 2b) sensorily detect spatially spaced focal planes (5) when the focusing optics are detuned to a certain extent. [2] Image recording method according to claim 1, wherein during the autofocus, by adjusting the focusing optics (4), a position of a focal plane (5) of the imaging optics (3) is shifted in a z-scan along an optical z-axis (6) in order to set / find a current best focal plane (7) which is to be used in the subsequent recording of the image data and / or - that the autofocus is carried out in order to determine a position z0 of a / the current best focus plane (7) and / or an adjustment of the focusing optics (4) which corresponds to a / the current best focus plane (7). [3] Image recording method according to one of the preceding claims, wherein during the autofocus a first image is recorded with the first image sensor (2a) at a time t1 and a second image is recorded with the second image sensor (2b) at a time t2, in particular while the focusing optics (4) are being adjusted, - preferably such that a respective position z1 and z2 of the focal plane (5) differs at times t1 and t2. [4] Image recording method according to one of the preceding claims, wherein a time offset t2-t1 between successive measurement times t1 and t2, at which the first image and the second image are respectively recorded, is at least 10%, preferably more than 25%, of a time interval between two directly successive individual images of the first or second image sequence. [5] Image recording method according to one of the preceding claims, wherein, within the scope of the autofocus, a first image sequence (8a) is recorded with the first image sensor (2a) and a second image sequence (8b) is recorded with the second image sensor (2b), - preferably wherein, to determine the position z0 of the instantaneous best focus plane (7), both image sequences (8a, 8b) are evaluated together, in particular by comparing individual images of the two image sequences (8a, 8b). [6] Image recording method according to one of the preceding claims, wherein the recording of the first image sequence (8a) is triggered in time with a first synchronization signal (9a) and the recording of the second image sequence (8b) is triggered in time with a second synchronization signal (9b), and wherein - the first and the second synchronization signal (9a, 9b) show a time offset and / or wherein the second synchronization signal (9b) is generated by means of a phase shift of a jointly used synchronization signal (9a). [7] Image recording method according to one of the preceding claims, wherein image data recorded during the autofocus with the first image sensor (2a) and / or with the second image sensor (2b) are first adapted via an image processing algorithm before the image data thus adapted are evaluated within the scope of the autofocus, - in particular wherein the image data are adapted with regard to an optical perspective and / or processed in such a way that a comparison of the first and the second image is simplified. [8] Image recording method according to one of the preceding claims, wherein an adjustment speed of the focusing optics (4) is changed during the autofocus, - preferably wherein the adjustment speed of the focusing optics (4) is slowed down when passing through the z-scan when approaching the current best focus plane (7). [9] Image recording method according to one of the preceding claims, wherein the method is carried out fully automatically by a controller of the visualization system (1), in particular without user input, - in particular as soon as the controller independently detects that the current image quality of the image data just recorded with the visualization system (1) no longer meets the specified requirements. [10] Image recording method according to one of the preceding claims, wherein the first image sensor (2a) and the second image sensor (2b) are realized by two non-overlapping, preferably juxtaposed, regions (13a, 13b) of an active sensor surface (12) of a single image sensor chip (16) of the visualization system (1), and wherein the two regions (13a, 13b) of the single image sensor chip (16) are read out with a temporal offset during the autofocus. [11] Image recording method according to one of the preceding claims, wherein the two image sensors (2a, 2b) are used to - to take stereoscopic images and / or - to detect spectrally different wavelength ranges by sensors, in particular wherein for this purpose an imaging beam path of the imaging optics (3) is divided by means of a spectrally selective beam splitter (10) into a first imaging beam path (11a) detected by sensors with the first image sensor (2a) and into a second imaging beam path (11b) detected spatially separately with the second image sensor (2b). [12] Image recording method, in particular according to claim 1, wherein image data, in particular in the form of a video image data stream, are recorded in an image recording mode with at least one image sensor (2a, 2b) of a visualization system (1) which has an imaging optics (3) with a focusing optics (4), and - wherein before recording the image data in image recording mode, an autofocus is carried out by detuning the focusing optics (4), characterized by , - that during the autofocus, the at least one image sensor (2a, 2b) is read out at different measuring times and that at each of these measuring times, less image data is read out and used for evaluation than is read out for individual images in the subsequent image recording mode and / or - that the at least one image sensor (2a, 2b) is operated during the autofocus at a first refresh rate which is higher than a second refresh rate at which the at least one image sensor (2a, 2b) is operated in the subsequent image recording mode. [13] Image recording method according to claim 12, - wherein a respective image information which is captured by the at least one image sensor (2a, 2b) at the different measuring times is a reduced image information which is obtained by line-by-line and / or column-by-column and / or block-by-block, in particular alternating, reading and non-reading of pixels of the at least one image sensor (2a, 2b) and / or - wherein a respective piece of image information which is acquired with the at least one image sensor (2a, 2b) at the different measuring times is acquired with the respective image sensor (2a, 2b) on the basis of a "binning mode", in which "binning mode" adjacent image areas or individual pixels are combined in terms of signal technology. [14] Visualization system (1), in particular comprising an endoscope or a microscope, with a first image sensor (2a), a second image sensor (2b), an imaging optics (3) with a tunable focusing optics (4), an image processing unit (14) and a controller (15), - wherein the controller (15) is configured to record respective image data, in particular in the form of a respective video image data stream, in particular in parallel / simultaneously with each of the two image sensors (2a, 2b) in an image recording mode and, if necessary, to perform an autofocus by detuning the focusing optics (4) before recording the image data in an autofocus mode, characterized by , - that the controller (15) is further configured to carry out a method according to one of the preceding claims. [15] Visualization system (1) according to the preceding claim, wherein the two image sensors (2a, 2b) - are spatially separated from each other or - are formed by a respective region (13a, 13b) of an active sensor surface (12) of a single image sensor chip (16) and / or - sensor-detect a respective imaging beam path (11a, 11b) which is generated with the common imaging optics (3) and / or - detect spectrally different imaging light, in particular after splitting by means of a spectrally selective beam splitter (10), by sensors and / or - offer different spectral sensitivity and / or spatial resolution and / or refresh rates.

Images