Visual object, method and computer program product for determining one or more visual properties of a test subject

Abstract

The invention relates to a visual object (100) and to a method and a computer program product for determining one or more visual properties of a test subject. The visual object (100) comprises at least two different filtered images which are combined to form a test image, wherein each of the filtered images represents one of at least two predefined different source images and is provided with a specific filter for a predefined optically induced blur.

Description

[0001] Visual object, method and computer program product for determining one or more visual properties of a subject

[0002] Description

[0003] The invention relates to a visual object, as well as a method and a computer program product for determining one or more visual characteristics of a subject. In particular, the invention can be assigned to the field of optometry.

[0004] To determine the visual characteristics, i.e., the visual acuity and / or certain visual functions (such as visual acuity, contrast perception, binocular vision, spatial and temporal resolution, gaze-motor skills, accommodation, adaptation, dim vision, day vision, etc.) of a test person or subject, measurement methods that utilize visual objects are used. For example, to determine visual acuity, the subject is presented with periodic patterns or standardized images that they are asked to recognize and identify on a test surface (such as a blackboard, screen, or projection screen).

[0005] Traditionally, optotypes are primarily used. Such optotypes are usually graphic symbols (e.g., numbers, letters, and / or figures) that are used, for example, to determine visual acuity. Well-known optotypes include the Snellen letter, the Snellen hook, and the Landolt ring. The key to an optotype is that it can be clearly identified or named by the test subjects.

[0006] If, for example, during an optometric examination, optotypes are to be recognized in the presence of optically induced blur (e.g. due to defocus and / or astigmatism and / or higher-order aberrations), this is often difficult as the blur in at least one eye of the test subject increases, since the test subject cannot assign the optotype seen blurred to an easily identifiable object.

[0007] Typically, a subject recognizes common optotypes (e.g., Landolt rings, numbers, or letters) less accurately when viewed with optically induced blur than without. A subject with optically induced blur may therefore take a long time to classify an optotype.

[0008] EP 2 243 419 B1 describes visual objects or optotypes whose average luminance is essentially equal to the luminance of the background. When viewed with optically induced blur, the optotypes disclosed in EP 2 243 419 B1 are either seen or not seen as a whole, depending on the fineness of the lines. The test subject must report which or how many objects were seen. It is therefore not sufficient to name a single optotype, which can lead to errors or inaccuracies in the determination of visual properties.

[0009] It is therefore an object of the present invention to provide a visual object and a method for the simple, reliable, and rapid determination of a subject's visual characteristics. This object is achieved by the subject matter of the independent claims. Advantageous embodiments are the subject matter of the dependent claims.

[0010] A first independent aspect for solving the problem relates to a visual object for determining one or more visual properties of a test subject, comprising at least two different filtered images combined (in particular superimposed) to form a test image, wherein each of the filtered images represents one of at least two predetermined different output images and is provided with a filter specific for a predetermined optically induced blur. In particular, each filtered image is assigned to an output image and a filter or a filter function. In particular, each of the filtered images represents one of at least two predetermined different output images. In particular, each of the filtered images is provided with a filter belonging to or assigned to its represented output image and specific for a predetermined optically induced blur.

[0011] The visual object is, in particular, represented by conventional means, including a projection onto the retina of a subject (in particular, by means suitable for determining refraction, such as refractive or diffractive lenses and a screen). In other words, the representation of the visual object can be achieved by technically common methods, such as, for example, appropriately designed objects (in particular, a printout of the visual object, e.g., on paper), front-lit or backlit panels, projection onto a board or wall, displays on a screen or monitor, and projection directly onto the retina of at least one eye of the subject.

[0012] The term "visual properties" of a subject particularly encompasses the subject's visual acuity and / or various visual functions. For example, with the help of one or more visual objects according to the invention, values ​​or thresholds for visual acuity, contrast perception or contrast sensitivity, binocular and stereo vision, adaptability or adaptation speed, etc., can be determined. In particular, the visual object and the method described herein serve to assess and / or optimize the subject's visual acuity.

[0013] A “test image” is understood in particular to be the overall image which combines and / or superimposes the individual filtered images. The filtered images are based on at least two predefined, different source images. In particular, more than two (e.g., three, four, five, six, etc.) source images or a plurality of source images can be predefined. In the context of this invention, an “source image” is referred to in particular as an original, unfiltered image. A source image can, for example, be one or more symbols, optotypes, graphics, or other objects that are easily and clearly recognizable for a test subject. In this context, “objects” are understood to mean not only symbols or objects, but also, for example, plants, animals, people, or figures.For example, a source image can include or contain a photograph or an image of an object familiar to the subject, a person known to the subject, or a cartoon character known to the subject. The given source images differ from one another, i.e., each of the given source images includes or shows a different, clearly recognizable or identifiable object.

[0014] Each filtered image represents an original image, meaning that each filtered image is assigned an original image (from the predefined original images). Each filtered image is associated with one of the predefined original images. In other words, the original image represented by a filtered image is therefore an original image associated with the filtered image.

[0015] Each of the filtered images is provided with a filter or filter function. This means that the source image represented by the filtered image or associated with the filtered image is filtered with a respective filter or filter function to generate the filtered image. In other words, the corresponding filter or filter function is applied to the source image.

[0016] A filter or a filter function is specific for a given (in particular predetermined and / or predefined) optically induced blur. In the context of this invention, “optically induced blur” is to be understood in particular as the blur of the image of an object (e.g. the visual object or the test image) on the retina of at least one eye of the test subject. When imaging an object on the retina of at least one eye of the test subject, the entire imaging process is taken into account, i.e. the imaging through the at least one eye of the test subject and, if applicable, additionally the imaging through any further optical system (e.g. refractive glasses, a contact lens, a phoropter, and / or other lenses), which, depending on the application, may be arranged in front of the at least one eye of the test subject (i.e. between the at least one eye of the test subject and the object).If no further optical system is arranged in front of the at least one eye of the subject, only the image through the at least one eye of the subject is taken into account. The entire imaging process, which creates the “optically induced blur” or the blur of the image of the object on the retina of the at least one eye of the subject, therefore takes into account any existing (i.e. any) ametropia of the at least one eye of the subject as well as any existing (i.e. any) additional optical system (e.g. refractive glasses, contact lenses, phoropters, and / or other lenses) which is arranged in front of the at least one eye of the subject. Furthermore, the pupil size of the at least one eye of the subject can also be taken into account in the imaging process.In particular, the “optically induced blur” of the subject is based on any existing ametropia of at least one eye of the subject, the pupil size of at least one eye of the subject, and possibly on any additional optical system arranged in front of at least one eye of the subject. The “optically induced blur” can include or be an intrinsic blur of at least one eye of the subject (referred to in the context of the invention as “intrinsically induced blur”) and / or an extrinsic optically induced blur (referred to in the context of the invention as “extrinsically induced blur”), which is caused or induced, for example, by an additional optical system arranged in front of at least one eye of the subject (e.g., refractive glasses, contact lenses, phoropters, and / or other lenses).If there is no intrinsic blur of at least one eye of the subject, the “optically induced blur” can, for example, only comprise or be an extrinsically induced blur of an otherwise emmetropic eye. If no further optical system is arranged in front of the at least one eye of the subject, the “optically induced blur” can, for example, only comprise or be the intrinsic blur of at least one eye of the subject. However, the “optically induced blur” can also be a combination (e.g., a sum) of an intrinsically induced blur and an extrinsically induced blur. It is noted that the “optically induced blur” is generally not causally dependent on the “intrinsically induced blur” or an “extrinsically induced blur.”

[0017] If the test images (or the visual object) are used to determine the visual impairment of at least one eye of the test subject, it can be provided that the test subject views the test images without an additional optical system (i.e. without glasses, contact lenses, phoropter, etc.). In this case, the "optically induced blur" corresponds to the intrinsic blur of at least one eye of the test subject, i.e. the "intrinsically induced blur". In another application, in which, for example, the "sensitivity to optically induced blur" based on full correction is to be measured, it can be provided that the test subject views a test image through an additional optical system (e.g. glasses, contact lenses, phoropter, etc.). In this case, the "optically induced blur" is a combination of any intrinsically induced blur and extrinsically induced blur.The additional optical system can be designed in such a way that the refractive error in at least one of the test subject's eyes is only partially corrected. A partial correction can result in what is known as fogging.

[0018] Each filter (or filter function) is associated with a specific optically induced blur. In particular, each filter (or filter function) is defined as a function of an optically induced blur. Each source image can thus have one or more associated filters or filter functions. Accordingly, each source image can have one or more associated filtered images.

[0019] In particular, a filter or a filter function is specific to a given optically induced blur in that a viewer (test subject), whose at least one eye, in combination with the effect of an optical system (e.g. refractive glasses), has a given optically induced blur, can just barely correctly assign an output image filtered with the filter function assigned to this optically induced blur. The expression “just barely correctly assign” can be understood or defined, for example, as the threshold of a psychometric function, wherein the recognizability of the output image assigned to the filtered image or the probability for the correct assignment of the filtered image to the output image preferably changes significantly when the effect of the optical system arranged in front of at least one eye of the test subject changes. This can, for example,be determined experimentally by determining the probability of correctly assigning the filtered image to the original image as a function of the optically induced blur for one subject or for a large number of subjects using psychometric methods.

[0020] Applying a filter associated with, or specific to, an optically induced blur to an original image can ensure the original image's recognizability in the filtered image created by applying the filter, when viewed through the optically induced blur. A filter or filter function can, for example, correspond to a simple convolution with a Gaussian kernel (an example of a low-pass filter) or the application of other filters such as high-pass or band-pass filters.

[0021] In order to assign a filter function or a filtered output image to an optically induced blur, the filter function can also be calculated from the given optically induced blur and the eye biometrics. For example, the spatial frequency edge of a low-pass filter can be selected such that it is proportional to the spatial frequency edge of the moment transfer function in the presence of optically induced blur (with a proportionality factor of 1, for example, both edges would be identical). The latter can also be approximately calculated as the reciprocal of the viewing angle of the scattering disk for a given optically induced blur. The viewing angles can be calculated, for example, as described in WO 2019034525 A1.

[0022] For example, each filtered image belonging to the test image can represent a different source image. However, it is also possible for the test image to comprise multiple filtered images of the same source image, with each of these filtered images being associated with a different optically induced blur.

[0023] The observation of an object with an “optically induced blur” means in particular in the context of the present invention that a test subject with an optically induced blur can only recognize or perceive the observed object in a blurred manner or only to a certain degree. The “optically induced blur” can be described mathematically, for example, with the help of the power vector P which is commonly used in spectacle optics. The power vector comprises the three components M, J o and J 45, where M denotes the spherical equivalent and J o, J 45 the astigmatic components. A power vector of P = (0,0,0) therefore means that an object is observed without optical blur, whereas a power vector of P = (M,0,0) with M 0 means that the object is observed with a certain optical blur, which in this case is quantified by M. As already mentioned, the “optically induced blur” can be specified or quantified, for example, with the help of a power vector.quantified. Thus, a filter or filter function used within the scope of the present invention, which is specific for a given optically induced blur, can additionally or alternatively be assigned a corresponding power vector. In addition to the deviation of the correction arranged in front of the eye (e.g. expressed as a power vector), the optically induced blur can also depend on the pupil size. Therefore, both a power vector and a pupil size can be used to quantify the optically induced blur. Corresponding filters can be used as a convolution of a point spread function, in particular of an uncorrected or incompletely corrected eye, with the original image to simulate an image with optically induced blur, as described, for example, in D. Kordek, LK Young and J.Kremlácek: “Comparison between optical and digital blur using near visual acuity”, Scientific Reports 11 , 3437 (2021 ), DOI 10.1038 / s41598-021 -82965-z. The quantification of an optically induced blur only via a power vector without explicitly specifying a pupil size can be carried out, for example, by assuming an average pupil size (e.g. 1.5 mm radius), and / or using a pupil calculated or predicted from other parameters (see WO 2013 087212 A1 ), and / or - particularly preferably - using an individually measured pupil.

[0024] Within the scope of the present invention, it was recognized that it is helpful to use test images which are designed in such a way that, with different optically induced blur, they each represent different objects which are easily recognizable by the test subject and which can be named simply and reliably. With the aid of the present invention, it is possible, for example, for the person being examined or the test subject to recognize an object A without optically induced blur (power vector P = (0, 0, 0)), but an object B in the presence of an optically induced blur which is not equal to zero (e.g. power vector P = (M, 0, 0)). With the visual objects or test images according to the invention, it is particularly possible for the test subject to recognize a specific initial image (from the plurality of predetermined initial images) depending on the optically induced blur, namely the initial image which was filtered with the filter specific for the corresponding optically induced blur.This advantageously allows the tested person to make a faster and more precise decision regarding a specific object or source image. With the help of the present invention, the determination of the test subject's visual characteristics can thus be carried out more easily, reliably, and quickly.

[0025] In a preferred embodiment, each of the filtered images is provided with a filter associated with its represented output image and specific for a predetermined optically induced blur such that on the test image, when viewed with a first predetermined optically induced blur, a first of the at least two output images is better recognizable than each of the remaining output images which differs from the first output image (or which does not correspond to the second output image), and when viewed with a second predetermined optically induced blur, a second of the at least two output images is better recognizable than each of the remaining output images which differs from the second output image (or which does not correspond to the second output image).

[0026] "When viewing (the visual object or the test image) with a predetermined optically induced blur" should in particular mean that the viewing takes place with otherwise identical viewing conditions (e.g., same brightness, same distance) and otherwise identical visual properties (e.g., with otherwise ideal visual properties or with otherwise average or standard visual properties). The term "otherwise" refers here to all visual properties with the exception of the optically induced blur. In other words, viewing with a first predetermined optically induced blur differs from viewing with a second predetermined optically induced blur solely in the different optically induced blur itself, but not in other parameters (in particular, parameters of at least one eye).

[0027] In particular, when viewed with the first predetermined optically induced blur, essentially only the first of the at least two original images is recognizable, while the remaining output images (which differ from the first original image) (e.g. the second original image) can preferably only be perceived as a blurred and / or unclear background. An unclear background can, for example, be an image which is perceived with a low or insufficient contrast. Accordingly, in particular when viewed with the second predetermined optically induced blur, essentially only the second of the at least two original images is recognizable, while the remaining output images (which differ from the second original image) (e.g. the first original image) can preferably only be perceived as a blurred background.

[0028] In particular, a filtered image is essentially only recognizable as the respective corresponding original image when viewed with a predetermined optically induced blur associated with the original image of the filtered image. In particular, this filtered image is essentially perceived as a blurred background when viewed with an optically induced blur that deviates from the predetermined optically induced blur associated with the original image of the filtered image, in particular by a certain value.

[0029] In a further preferred embodiment, each of the filtered images is provided with a filter associated with its original image and specific for a predetermined optically induced blur such that the filtered image, when viewed with the predetermined specific optically induced blur, has a predetermined similarity and / or a predetermined recognizability, in particular a predetermined minimum similarity and / or a predetermined minimum recognizability, with its associated original image. Preferably, when viewed with the predetermined specific optically induced blur, the filtered image has a similarity of (at least) 50%, more preferably (at least) 60%, even more preferably (at least) 70%, even more preferably (at least) 80%, even more preferably (at least) 90%, even more preferably (at least) 95%, and most preferably (at least) 99%.The values ​​given above can, in particular, be used to determine recognition probabilities in a psychometric test. However, if other similarity measures are used, other values ​​may also be advantageous, which, for example, do not necessarily have to be between 0% and 100%.

[0030] In particular, the filter (or the corresponding filter function) associated with a specific output image and specified for a specific optically induced blur is characterized in that the associated output image filtered with this filter (or with this filter function) has a predetermined, in particular minimal, similarity to the associated output image when viewed with the associated or specific optically induced blur.

[0031] In a further preferred embodiment, the at least two filtered images comprise a first filtered image F1(B1) and a second filtered image F2(B2).

[0032] The first filtered image F1(B1) represents a first output image B1 filtered with a first filter or a first filter function F1. The second filtered image F2(B2) represents a second output image B2 filtered with a second filter or a second filter function F2, which differs from the first output image B1. In particular, a similarity S 111 of the first filtered image F1(B1) with its original image Bi when viewed with a first optically induced blur P1 greater than a similarity S 212 of the second filtered image F2(B2) observed at the first optically induced blur P1 with its original image B2. Thus, S 111 > S 212 . In general, S abc the similarity of the a-th filter function F a filtered a-th output image B a , i.e. from F a (B a ), with the c-th output image B c, when viewed with the b-th optically induced blur P b If the optically induced blur is considered as a function, it can be written

[0033] S abc = S(P b (F a (B a )),B c ), where S(., . ) denotes any similarity function of two images, and P b (. ) the effect of optically induced blur P b on a picture.

[0034] In particular, for i ∈ Z + ; k ∈ Z + ; i ≤ N; i ≤ M; k ≤ N; and i ≠ k:

[0035] S iii > S kik , where S iii a similarity of an i-th filtered image F i (B i ) with its original image B i when viewed with an i-th optically induced blur P i is, and where S kik a similarity of a k-th filtered image Fk(Bk) with its source image B kwhen viewed with the i-th optically induced blur P i Where F i (B i ) the i-th output image B filtered with an i-th filter function Fi i Accordingly, Fk(Bk) denotes the kth output image Bk filtered with a kth filter function Fk. Furthermore, N denotes the number of input images used, and M denotes the number of predefined, different optically induced blurs. N does not necessarily have to equal M.

[0036] The similarity of two images can be determined or defined, for example, using image metrics to determine differences between the two images. Known methods or metrics such as cross-entropy, Kullback-Leibler divergence, Earth-mover's metric, etc. can be used for this purpose. Further examples of image metrics that can also be used within the scope of the present invention to define the similarity of two images are given or described, for example, in US6493023B1. The similarity can be defined, for example, as the negative or reciprocal of the image metric representing a difference.

[0037] Alternatively or additionally, the similarity can be determined or defined using a physiologically motivated model. This can be done, for example, with the help of a so-called "spatial standard observer," which is described in the patents WO2006 / 079115A2 and US7783130B2. In particular, these documents describe the determination of an image metric ("visibility metric"), which can be used to define the similarity of two images within the scope of the present invention. Further information on physiologically motivated models and / or image metrics that can be used within the scope of the present invention can be found, for example, in the following publications:

[0038] - AB Watson et al: “A standard model for foveal detection of spatial contrast”, Journal of Vision (2005) 5, pages 717-740, DOI: 10.1167 / 5.9.6;

[0039] - A. B. Watson et al: „Predicting visual acuity from wavefront aberrations“, Journal of Vision (2008) 8 (4): 17, Seiten 1 -19, DOI: 10.1167 / 8.4.17;

[0040] - A. B. Watson et al: „Blur clarified: A review and synthesis of blur discrimination“, Journal of Vision (2011 ) 11 (5): 10, Seiten 1 -23, DOI: 10.1167 / 11.5.10;

[0041] - A. B. Watson et al: „Modeling acuity for optotypes varying in complexity“, Journal of Vision (2012) 12(10): 19, Seiten 1 -19, DOI: 10.11167 / 12.10.19.

[0042] The disclosure content of all the above-mentioned patents and publications is hereby expressly incorporated by reference. Since the determination or definition of image metrics as such is familiar to those skilled in the art, it will not be discussed further within the scope of the present invention. A concrete embodiment for the definition or determination of a similarity or a similarity function is described below in connection with exemplary embodiments of the present invention.

[0043] Alternatively or additionally, the similarity of two images can also be determined or defined based on a recognizability, whereby the recognizability is determined by a model based on data from a test with test subjects.

[0044] Alternatively or additionally, the similarity between two images can be quantified by calculating the classification probability of a classifier (e.g., a neural network or other machine-learning model), where the classifier has previously been trained with an image dataset containing at least the source images. A specific embodiment of a classifier is described below in connection with exemplary embodiments of the present invention. However, training must not be performed with blurred images.

[0045] In a further preferred embodiment, the following applies (in particular, the filters or filter functions are selected such that the following applies): S 111 > S 121 ; and / or

[0046] S 222 > S 121 ; and / or

[0047] S 222 > S 212 ; where S 111is the similarity of the first filtered image F1(B1) considered at the first blur P1 with its original image B1, where S 222 is the similarity of the second filtered image F2(B2) considered at a second blur P2 with its original image B2, where S 121 is the similarity of the first filtered image F1(B1) considered at the second blur P2 with its original image B1, and where S 212 the similarity of the second filtered image F2(B2) considered at the first blur P1 with its original image B2.

[0048] The relationship S 111 > S 121 In words means that the similarity S 111 of the first filtered image F1(B1) considered at the first blur P1 with its original image B1 is greater than the similarity S 121 of the first filtered image F1(B1) with its original image B1 when viewed with a second blur P2.

[0049] In particular, for i ∈ Z+ ; k ∈ Z + ; i ≤ uN; i ≤ M; k ≤ N; k ≤ M; and i # k:

[0050] S iii > S iki , where S iii a similarity of an i-th filtered image F i (B i ) with its original image Bi when viewed with an i-th blur P i is, and where S iki the similarity of an i-th filtered image F i (B i ) with its original image Bi when viewed with a k-th blur Pk. Here, F denotes i (B i ) which is a I-th filter function F i filtered i-th output image B i . Furthermore, N denotes the number of given initial images and M the number of given different optically induced blurs. N does not necessarily have to equal M. The relationship S 222 > S 121 means in words that a similarity S 222of the second filtered image F2(B2) viewed at a second blur P2 with its original image B2 is greater than a similarity S 121 of the first filtered image F1(B1) viewed at the second blur P2 with its original image B1.

[0051] In particular, for i ∈ Z + ; k ∈ Z + ; i ≤ N; k ≤ N; k ≤ M and i + k:

[0052] S kkk > S iki , where S kkk is the similarity of the k-th filtered image Fk(Bk) considered at a k-th blur Pk with the corresponding k-th original image Bk, and where S iki the similarity of the i-th filtered image F considered at the k-th blur Pk i (B i ) with the corresponding i-th output image Bi. Here, N denotes the number of given output images and M the number of given different optically induced blurs. N does not necessarily have to equal M.

[0053] The relationship S 222 > S 212 means in words that a similarity S 222 of the second filtered image F2(B2) viewed at a second blur P2 with its original image B2 is greater than a similarity S 212 of the second filtered image F2(B2) with its original image B2 when viewed with the first blur P1.

[0054] In particular, for i ∈ Z + ; k ∈ Z + ; i ≤ M; k ≤ N; k < M; and i # k:

[0055] S kkk > S kik . where S kkk is the similarity of the k-th filtered image Fk(Bk) considered at a k-th blur Pk with the corresponding k-th original image Bk, and where S kik the similarity of the i-th blur P i considered k-th filtered image Fk(Bk) with the k-th original image B kWhere N denotes the number of given initial images and M denotes the number of given different optically induced blurs. N does not necessarily have to equal M.

[0056] In a further preferred embodiment, the following applies (in particular, the filters or filter functions are selected such that the following applies):

[0057] S 112 < S 111 and S 112 < S 222 ; and / or Si22 < S 111 and S122 < S 222 ; where S 111 is the similarity of the first filtered image F1(B1) considered at the first blur P1 with its original image B1, where S 222 is the similarity of the second filtered image F2(B2) considered at a second blur P2 with its original image B2, where S 112 the similarity of the first filtered image F considered at the first blur P1 i(B1) with the original image B2, and where S122 is the similarity of the first filtered image F1(B1) considered at the second blur P2 with the original image B2.

[0058] The relationships S 112 < S 111 and S 112 < S 222 mean in words that a similarity S 112 of the first filtered image F1(B1) with the second original image B2when viewed with the first blur P1is both smaller than the similarity S 111 of the first filtered image F1(B1) considered at the first blur P1 with its original image B1 , and is smaller than a similarity S 222 of the second filtered image F2(B2) considered at the second blur P2 with its original image B2.

[0059] In particular, for i ∈ Z + ; k ∈ Z + ; i ≤ N; i ≤ M; k ≤ N; k ≤ M; and i # k:

[0060] S iik < S iii And Siik S kkk, where Siü is the similarity of the i-th uncertainty P i considered i-th filtered image F i (B i ) with the corresponding i-th output image Bi, where S kkk is the similarity of the k-th filtered image Fk(Bk) considered at a k-th blur Pk with the corresponding k-th output image Bk, and where Siik is the similarity of the image Fk(Bk) considered at an i-th blur P i considered i-th filtered image F i (B i ) with the kth output image Bk. Here, N denotes the number of given output images and M the number of given different optically induced blurs. N does not necessarily have to equal M.

[0061] The relationships S122 < S 111 and S122 < S 222mean in words that the similarity S122 of the first filtered image F1(B1) with the second original image B2 when viewed with a second blur P2 is both smaller than the similarity S111 of the first filtered image F1(B1) viewed with the first blur P1 with its original image B1, and is also smaller than the similarity S 222 of the second filtered image F2(B2) considered at the second blur P2 with its original image B2.

[0062] In particular, for i ∈ Z + ; k ∈ Z + ; i ≤ N; i ≤ M; k ≤ N; k ≤ M; and i # k:

[0063] S ikk < S iii And S ikk < S kkk , where S iii the similarity of the i-th blur P i considered i-th filtered image F i (B i ) with the corresponding i-th output image Bi, where S kkkis the similarity of the k-th filtered image Fk(Bk) considered at a k-th blur Pk with the corresponding k-th original image Bk, and where S ikk the similarity of the i-th filtered image F considered at a k-th blur Pk i (B i ) with the kth output image Bk. Here, N denotes the number of given output images and M the number of given different optically induced blurs. N does not necessarily have to equal M.

[0064] In a further preferred embodiment, the at least two filtered images are arranged next to each other and / or overlapping, in particular at least partially superimposed. In particular, the filtered images of the test image can be arranged next to each other, e.g., in a row, a column, a matrix, or otherwise non-overlapping. Alternatively or additionally, the filtered images of the test image can be arranged overlapping and in particular superimposed. In the latter case, the overlapping, in the case of blur P i images that do not appear sharp F j (B j ) with j + i a (blurred) background from which the clearly recognizable image F i (B i) contrasts with the original image Bi, making it easy for the observer or subject to decide whether to classify the visual impression as image Bi. At least some or all of the filtered images can be superimposed (e.g., added together). The test image created, e.g., by superimposing the individual filtered images, can then be displayed or presented to the subject using conventional means (e.g., the test image can be displayed on a screen or printed).

[0065] In a further preferred embodiment, the filters or filter functions are adapted to average higher-order aberrations of an eye. Alternatively, the filters or filter functions are adapted to individual higher-order aberrations of at least one eye of the subject. In this way, it can advantageously be achieved that the subject recognizes the output image specifically filtered for their optically induced blur better, i.e., more clearly and quickly. In other words, recognizability can be improved individually or at least on average.

[0066] Any aberrations can be taken into account, for example, with the help of a prefiltering technique. In the prefiltering technique, the original image is convolved with a function that has small values ​​where the point spread function (especially of an uncorrected or not fully corrected eye) has large values, and vice versa, e.g. with the reciprocal of the point spread function or with a Wiener filter based on the point spread function. Negative pixel intensities of an image prefiltered in this way can be normalized to a displayable range (e.g. relative intensities between 0 and 1). If an image prefiltered in this way is viewed through optically induced blur, the viewer can perceive the (sharp) original image, which, however, has a lower contrast due to the normalization. Further information on this can be found, for example,in Fu-Chung Huang: “A Computational Light Field Display for Correcting Visual Aberrations”, EECS Department, University of California, Berkeley, 2013, Technical Report No. UCB / EECS-2013-206, http: / / www.eecs.berkeley.edu / Pubs / TechRpts / 2013 / EECS-2013-206.html.

[0067] In particular, higher-order aberrations can now be accounted for by considering the phase error of higher (Zernike) order aberrations in addition to the second (Zernike) order phase error when calculating the point spread function from the phase error in the entrance pupil (often also called the "wavefront"). The calculation of the point spread function from the phase error in the entrance pupil is described, for example, in D. Kordek, LK Young, and J. Kremläcek: "Comparison between optical and digital blur using near visual acuity," Scientific Reports 11, 3437 (2021), DOI 10.1038 / s41598-021-82965-z.

[0068] A further independent aspect for solving the problem relates to a method for determining one or more visual properties of a subject, comprising: providing a visual object according to the invention, wherein the provision of the visual object particularly comprises displaying the visual object. The display can be carried out, in particular, using conventional means, including projection onto the retina of a subject, or by displaying it on a screen or by printing it out, e.g., on paper. In particular, the visual object can be presented to the subject.

[0069] In a preferred embodiment, providing a visual object comprises the following steps: - providing at least two output images; - generating at least two filtered recognition images by applying filters (or filter functions) specific for a predetermined optically induced blur to the at least two output images such that, when viewing the filtered recognition images with the predetermined optically induced blur, each filtered recognition image has a predetermined similarity, in particular a predetermined minimum similarity, with the respective output image associated with the filtered recognition image.

[0070] A “filtered recognition image” is in particular a filtered initial image which is filtered in such a way that the test subject can recognize it as the initial image when viewed with an optically induced blur.

[0071] Preferably, the method for providing the visual object according to the invention comprises the steps of: - adapting the filter function(s) to average higher-order aberrations of an eye; and / or - adapting the filter function(s) to individual higher-order aberrations of at least one eye of the subject.

[0072] In a further preferred embodiment, providing the visual object comprises determining the filter (or a corresponding filter function) specific for a given optically induced blur for each output image. Determining the filter specific for a given optically induced blur for each output image comprises, in particular, calculating an optical transfer function associated with the respective optically induced blur. For example, the filtering can correspond to an at least partial phase reversal and / or phase rectification. Methods for this can be found, for example, in the publication by S. Ravikumar et al: "Phase changes induced by optical aberrations degrade letter and face acuity", Journal of Vision (2010) 10(14): 18, pages 1-12, DOI: 10.1167 / 10.14.18, the disclosure of which is hereby expressly incorporated by reference. For this purpose, it is necessary to determine the optical transfer function, i.e.the Fourier transform of the point spread function for the respective optically induced blur P. i to calculate.

[0073] Other filters can also be used. For example, to generate a filtered image, at least two filters from the filter categories low-pass filter, band-pass filter and high-pass filter can be used. When using low-pass, high-pass and / or band-pass filters, the characteristic frequencies of the filters can be selected depending on the characteristic frequencies of the optical transfer function generated by an optically induced blur. For example, when combining a low-pass or band-pass filter, which is applied to a first output image, with a high-pass filter, which is applied to a second output image, the filter edge of the low-pass filter (orthe high-frequency filter edge of the band-pass filter) lower than the spatial frequency of the filter edge of the optical transfer function generated by a given optically induced blur, and correspondingly the filter edge of the high-pass filter higher than the frequency of the filter edge of the optical transfer function.

[0074] An optimization procedure is also possible in which filter parameters are adjusted (e.g., the above-mentioned characteristic frequencies of the filters, or the filter parameters used in the "prefiltering technique"). An optimization function G0(. ) to be maximized can, for example, be determined from the similarities S ijk and depends implicitly on the parameters θ i the i-th filter functions F i (.) away: So the maximum of G θ formed via the parameters 0 of the filter functions: θ* = argmax θ (G θ ) .

[0075] It should be noted that several filter functions can be identical. For example, there may be only two types of filter functions, e.g., a high-pass filter and a band-pass filter, whose parameters are optimized. In this case (as in the example), the filter functions would correspond to, for example, F1(θ1) = F3(θ3) = F 2i-1 (θ 2i-1 ) = = F W-1 (θ W-1 ) = F BP (θ BP ) one and the same

[0076] Bandpass filter, and the filter functions F2(θ2) = F4(θ4) = F 2i (θ 2i ) = ... = F W (θ W ) = F HP (θ HP ) one and the same high-pass filter. Three parameters f lt f2 and f3 of the filter functions are varied during optimization using conventional optimization methods to achieve a maximum of G g {f1,f2,f3) zu where θ 2i-1 = e BP = the lower and upper edges of the bandpass filter, and θ 2i = θ HP= f3 is the frequency edge of the high-pass filter. The weights w ijk are positive as usual and can in the simplest case all be chosen to be identical 1, or e.g. as follows:

[0077] In a further preferred embodiment, providing the visual object comprises the following steps:

[0078] Filtering a plurality N of (given) different output images Bi with 1 ≤ i ≤ N, for a plurality M of (given) optically induced blurs P j with 1 ≤ j ≤ M, where M and N are each a positive integer with M < N, and

[0079] Selecting the filtered images F j (B i ), whose similarity S ii to the original images Bi when viewed with an optically induced blur Pi exceeds a first threshold value (ie is as high as possible), and their similarity S ij to the other output images B jwith j ≠ i when viewed with an optically induced blur P ifalls below a second threshold value (i.e. is as low as possible). This can be understood as an optimization across all subsets with M images. In particular, the first threshold value is greater than or equal to the second threshold value. Preferably, therefore, from a large number of N source images, only those M images are selected which are particularly suitable as images within the scope of the invention (due to the selection, M < N must be selected). The reason for the selection is that the recognizability of the source images in the filtered images depends not only on the filter functions used, but also on the distribution of the information required for assignment across different spatial frequencies. For example, a first source image which can already be easily recognized based on the low-frequency components can be ideally suited to being filtered with a low-pass filter in order to be combined later with a high-pass filtered second image.However, if the filters are swapped and the first output image is treated with a high-pass filter and the second output image with a low-pass filter, it may be that the high-frequency image components of the first output image are no longer suitable (or no longer as good) for recognizing the output image in the filtered image, because, for example, they are too unspecific or their amplitudes were already too low in the first output image.

[0080] In a further preferred embodiment, the selection of filtered images represents an optimization over all subsets with R output images. The optimization is preferably carried out by maximizing an objective function, wherein the objective function is defined in particular as follows: Where δ represents the Kronecker symbol and w ijk Weights are marked with w ijk ≥ 0 and In particular, only subsets of the original images β with a constant number of elements, with the filter functions now fixed. An optimization function G to be maximized β can be deduced from the similarities S ijk and depends implicitly on the parameters of the i-th filter functions F i ( . ) away:

[0081] The most suitable set of images can be determined by β* = argmax β∈T (G β ), where T is the set of all subsets of the indices of the original images with R elements. The weights w ijk can all be the same (e.g. = 1 for all i, j, k). However, it is advantageous to set the weights with the

[0082] Scale similarities of the diagonal elements as follows: for arbitrary k.

[0083] In a further preferred embodiment, the method comprises the steps of: evaluating a reaction of the subject to the visual object, wherein the evaluation is carried out in particular

[0084] - determining an ametropia of the subject, and / or

[0085] - determining the subject’s sensitivity to optically induced blur, and / or

[0086] - determining an optimal filter (or an optimal filter function) for the test subject.

[0087] A subject's reaction can be, for example, a response, a motor action (such as a demonstration, tapping images on a board, device, or screen), and / or an eye movement. Determining a subject's ametropia can thus advantageously be performed without having to create optically induced blur (using additional aids).

[0088] Determining an optimal filter for the test subject may comprise determining a filtering with a fixed but unknown optically induced blur, using the test images according to the invention, with the aim of finding a filtering that enables the best possible recognizability of an image filtered with the filter on a screen.

[0089] A further independent aspect for achieving the object relates to a computer program product comprising machine-readable program code which, when loaded onto a computer, is suitable for executing the inventive method described above. In particular, a computer program product is understood to mean a program stored on a data carrier. In particular, the program code is stored on a data carrier. In other words, the computer program product comprises computer-readable instructions which, when loaded into a computer memory and executed by the computer, cause the computer to execute an inventive method described above.

[0090] In particular, the computer program product may comprise a computer-readable storage medium having code stored thereon, wherein the code, when executed by a processor, causes the processor to implement a method according to the invention. The invention thus provides a computer program product, in particular in the form of a storage medium or a data stream, containing program code that, when loaded and executed on a computer, is designed to carry out a method for providing a visual object according to the invention and / or for determining one or more visual properties of a subject.

[0091] The above-mentioned further independent aspects and in particular the related preferred embodiments are also subject to the above-mentioned or below statements regarding the embodiments of the first aspect. In particular, the above-mentioned and below statements regarding the embodiments of the respective other independent aspects also apply to an independent aspect of the present invention and the related preferred embodiments.

[0092] In the following, individual embodiments for achieving the object are described by way of example with reference to the figures. In some cases, the individual embodiments described have features that are not absolutely necessary to carry out the claimed subject matter, but which provide desired properties in certain applications. Thus, embodiments that do not have all the features of the embodiments described below are to be regarded as falling within the scope of the described technical teaching. Furthermore, to avoid unnecessary repetition, certain features are only mentioned in relation to individual embodiments described below. It is pointed out that the individual embodiments should therefore not only be considered in isolation, but also in conjunction with one another.Based on this overview, those skilled in the art will recognize that individual embodiments may also be modified by incorporating one or more features of other embodiments. It is noted that a systematic combination of the individual embodiments with one or more features described with reference to other embodiments may be desirable and useful and should therefore be considered and considered to be encompassed by the description.

[0093] Short description of the drawings

[0094] Figure 1 a shows an exemplary first output image;

[0095] Figure 1 b shows an exemplary second output image;

[0096] Figure 2a shows an exemplary first filtered image, namely the first output image of Figure 1a filtered with a first filter;

[0097] Figure 2b shows an exemplary second filtered image, namely the second output image of Figure 1b filtered with a second filter;

[0098] Figure 3 shows a visual object or test image according to a preferred embodiment of the invention, in which the first filtered image of Figure 2a and the second filtered image of Figure 2b are superimposed;

[0099] Figure 4 shows: in the top row, six exemplary output images B1 to B6; in the middle row, masks for a high-frequency background associated with the respective output images, each of which was generated by applying a low-pass filter to the output images above; and in the bottom row, the output images of the top row, with a high-frequency background added in the area of ​​the masks (black in the middle row above).

[0100] Fig. 5 shows: in the upper row, the six exemplary output images of Figure 4, which were filtered with a bandpass filter; and in the lower row, the six exemplary output images of Figure 4, which were filtered with a highpass filter;

[0101] Fig. 6 shows combined filtered images based on the output images of Fig. 4 and the bandpass and highpass filtered output images of Fig. 5;

[0102] Fig. 7 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed without optical blur;

[0103] Fig. 8 shows: on the left the Zernike coefficients of a wavefront error of the at least one eye of a subject; top right, a 1024 x 1024 pixel image of the point spread function of the uncorrected eye with an entrance pupil radius of 2.52 mm for a wavelength of 550 nm; and bottom right, an enlargement of the overlying image.

[0104] Fig. 9 shows superimposed filtered images with different optically induced blurs;

[0105] Fig. 10 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed with a first optical blur P1; Fig. 11 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed with a second optical blur P2;

[0106] Fig. 12 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed with a third optical blur P3;

[0107] Fig. 13 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed with a fourth optical blur P4;

[0108] Fig. 14 shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6 when viewed with a fifth optical blur P5;

[0109] Fig. 15 shows a graphical representation of the similarities from the tables of Figures 7 and 10 to 14;

[0110] Detailed the

[0111] It is noted that the figures shown herein are photographic images, each of which cannot be represented as a black and white line drawing without undesirably distorting their character.

[0112] Figure 1a shows an example first source image. This first source image B1 shows a photo of "Kermit," a cartoon character from the well-known "Muppet Show." Figure 1b shows an example second source image. This second source image B2 shows a photo of "Miss Piggy," another cartoon character from the "Muppet Show." Both source images or photos are unfiltered and thus each correspond to an original image.

[0113] Figure 2a shows a first filtered image F1(B1) provided with a first filter or a first filter function Fi. This first filtered image F1(B1) is based on the first output image B1 or represents the first output image B1. In particular, the first filtered image F1(B1) represents the first output image B1 to which a first filter or a first filter function Fi has been applied. The first filter Fi is selected such that the first filtered image F1(B1) has a relatively high similarity to the first output image B1 when viewed with a first optically induced (spherical) blur P1, which corresponds, for example, to a first power vector P1 = (1.5 dpt, 0, 0). However, if the filtered image F1(B1) is provided with a second optically induced blur P2, which differs from the first optically induced blur P1 (in particular by a certain orminimum value), the first filtered image F1(B1) appears unclear or blurred and / or shows less similarity to the first original image than when viewed with the first optically induced blur, for which the first filter Fi is specific. As can be seen from Figure 2a, the filtered image F1(B1) shown therein appears unclear or blurred when viewed without optically induced (spherical) blur, which corresponds to a second power vector P2 = (0 dpt, 0, 0). The filter Fi is thus a filter specific for the first optically induced blur, since when viewed with the first optically induced blur, it allows the original image B1 to be clearly recognized, whereas when viewed with a different optically induced blur (for example, with the second optically induced blur P2), the original image B1 is less clearly, poorly, or not recognized at all (or only as a background).

[0114] Figure 2b shows a second filtered image F2(B2) provided with a second filter or a second filter function Fi. This second filtered image F2(B2) is based on the second output image B2 or represents the second output image B2. In particular, the second filtered image F2(B2) represents the second output image B2 to which a second filter or a second filter function F2 has been applied. The second filter F2 is selected such that the second filtered image F2(B2) exhibits a relatively high similarity to the second output image B2 when viewed with the second optically induced blur P2, which corresponds to the power vector P2 = (0 dpt, 0, 0). However, if the filtered image F2(B2) is viewed with a different optically induced blur that differs from the second optically induced blur P2 (in particular by a certain or minimal value), the filtered image F2(B2) appears unclear orblurred and / or exhibits less similarity to the second original image B2 than when viewed with the second optically induced blur P2, for which the second filter F2 is specific. As can be seen from Figure 2b, the filtered image F2(B2) shown therein appears clearly recognizable when viewed without optically induced (spherical) blur P2, which corresponds to the second power vector P2 = (0 dpt, 0, 0). The filter F2 is thus a filter specific for the second optically induced blur P2, since it allows the original image B2 to be clearly recognized when viewed with the second optically induced blur P2, whereas when viewed with a different optically induced blur (for example, with the first optically induced blur P1), the original image B2 is less clearly, poorly, or not recognizable at all (or only as a background).

[0115] In the illustrated embodiment, the filter Fi used is a low-pass filter, ie, the filtered image F1(B1) shown in Figure 2a is a low-pass filtered image. In contrast, the filter F2 used in the illustrated embodiment is a high-pass filter, ie, the filtered image F2(B2) shown in Figure 2b is a high-pass filtered image.

[0116] Figure 3 shows a visual object or a test image 100 according to a preferred embodiment of the invention. The first filtered image F1(B1) shown in Figure 2a and the second filtered image F2(B2) shown in Figure 2b are combined in the illustrated visual object 100 or test image. In the illustrated embodiment, the first filtered image F1(B1) of Figure 2a and the second filtered image F2(B2) of Figure 2b are superimposed. In other words, the first filtered image F1(B1) and the second filtered image F1(B1) completely overlap.

[0117] The visual object or test image 100, displayed with an edge length corresponding to 8 degrees at a proximity of 0 dpt, can be recognized as "Miss Piggy" (i.e., the second initial image B2) when viewed with the first spherical blur P2 = (0 dpt, 0, 0). If, on the other hand, the visual object or test image 100 is viewed with the second spherical blur P2 = (1.5 dpt, 0, 0), it can be recognized as "Kermit" (i.e., the first initial image B1). The information as to which initial image a test subject recognizes in the displayed test image can be used to determine or ascertain the test subject's visual characteristics (e.g., ametropia). For example, a plurality of visual objects or test images 100 according to the invention can be presented to the test subject in order to successively infer certain visual characteristics of the subject based on a corresponding reaction or response from the test subject.

[0118] Each of the filtered images F1(B1) and F2(B2) is thus provided with a filter associated with its represented original image B1 or F1(B1) and specific for a predetermined optically induced blur such that on the test image, when viewed with a first predetermined optically induced blur, a first of the at least two original images (here: "Kermit") is better recognizable than each of the remaining original images (here: "Miss Piggy"), wherein when viewed with a second predetermined optically induced blur, a second of the at least two original images (here: "Miss Piggy") is better recognizable than each of the remaining original images (here: "Kermit").

[0119] Within the scope of the present invention, a method for creating a test image from two or more source images, which contain, for example, symbols, optotypes, or other easily and clearly recognizable objects, is proposed. Each image Bi is filtered with a filter function Fi. The filtered image F i (B i ) has a similarity S for the optically induced blur Pj ij k with the original image Bk. At least some of the filtered images can then be superimposed (e.g., added together) and the image created by the superposition can be displayed on a screen or printed. The displayed filtered images can be arranged next to each other (e.g., in a row, a column, a matrix, or otherwise non-overlapping), and alternatively or additionally, they can also be arranged one above the other. In the latter case, the overlapping images, which are blurred P iimages that do not appear sharp F j (B j ) with j + i a background from which the clearly recognizable image F i (B i) stands out from the original image Bi, so that the viewing person can easily decide whether to classify the visual impression as image Bi. The visual objects or test images according to the invention are designed in particular in such a way that, with different levels of optically induced blur, they represent different objects that are easily recognizable by the test subject and can be named simply and reliably. For example, the person being tested can recognize object A without optically induced blur (as a power vector P = (0, 0, 0)), but object B with an optically induced blur that is different from zero (e.g. power vector P = (M, 0, 0)). With the test images according to the invention, a test subject can therefore recognize a different object in each case, particularly depending on the optically induced blur. As a result, the tested person decides in favor of a specific object more quickly and is clearer.

[0120] In order to generate the test images according to the invention, the images Bi can first be filtered in a suitable manner with the filter function Fi in order to achieve good recognizability with the desired optical blur P i The filtering can correspond to at least partial phase inversion and / or phase rectification. For this purpose, it is necessary to calculate the "Optical Transfer Function" (ie the Fourier transform of the "Point Spread Function") for the respective optically induced blur P i to calculate. Other filters can also be used. For example, to generate a filtered image, at least two filters from the filter categories of low-pass filter, band-pass filter, and high-pass filter can be used. The filters can also be adapted to the individual or average higher-order aberrations of the eye in order to improve detectability individually or at least on average.

[0121] Particularly good results are obtained by filtering a large number of different images Bi, with 1 ≤ i ≤ N, for the optically induced blurs Pj, with 1 ≤ j ≤ M, where M < N, and then filtering the filtered images F i (B i ) whose similarity S ii to the original images Bi when viewed with the blur P i as high as possible, and at the same time their similarity S ij to the other pictures B j with j # i when viewed with the blur P i is as low as possible. This can be understood as an optimization over all subsets with M images, where the optimization can be solved by maximizing an objective function. Possible applications of the invention include, in particular:

[0122] - the determination of ametropia without having to create optically induced blur;

[0123] - the determination of the sensitivity to optically induced blur; and / or

[0124] - the determination of a filtering in the case of fixed but unknown optically induced blur, in which the test images according to the invention are used, with the aim of finding a filtering which ensures the best possible recognizability of an image filtered with the filter on a screen.

[0125] When using test images, as with test images containing state-of-the-art optotypes, it is important to note that, in addition to the optically induced blur, the distance between the test image and the eye of the person being tested plays a significant role. It is also important to note that certain optometric parameters (e.g., visual acuity) determined with the test images according to the invention do not necessarily correspond to the same parameters as those determined with the aid of test images containing state-of-the-art optotypes, and that calibration of the measurement method according to the invention to a standard measurement method may be necessary.

[0126] In the following, further examples of source images, filtered images, combined or superimposed images and their similarity to the source images are presented using Figures 4 to 14.

[0127] Figure 4 shows six example output images B1 to B6 in the top row. Output image B1 represents an elephant, output image B2 a cat, output image B3 a person, output image B4 a moon, output image B5 an umbrella, and output image B6 a sun. The middle row of Figure 4 shows masks corresponding to the output images B1 to B6 shown in the row above, which were generated by applying a low-pass filter to the output images above them. These masks are used to provide the output images B1 to B6 with a high-frequency background. The bottom row of Figure 4 shows the output images B1 to B6 provided with a high-frequency background, with the high-frequency background being added by applying the masks to the output images.In particular, a high-frequency background was added to an image area defined by the masks (shown in black in the masks). The high-frequency background serves primarily to draw the viewer's attention to high-frequency structures in the combined image, provided they can be perceived with a given optically induced blur.

[0128] Figure 5 shows filtered output images. In the upper row of Figure 5, the output images B1 to B6 were each filtered with a bandpass filter Fi, while in the lower row of Figure 5, the output images B1 to B6 were each filtered with a highpass filter F2. For example, the upper left image labeled F1(B1) is the output image B1 filtered with the bandpass filter F1. The lower right image labeled F2(B6), for example, is the output image B6 filtered with the highpass filter F2.

[0129] Figure 6 shows a matrix with combined (or superimposed) filtered images. The images displayed in the matrix can be identified by two numbers, where the first number indicates the row and the second number the column of the matrix. The images contained in the diagonal of the matrix shown in Figure 6, i.e., images "11", "22", "33", "44", "55", and "66", correspond to the output images B1 to B6. The off-diagonal elements of the matrix contain images that represent a combination or superposition of bandpass and high-pass filtered output images. The same bandpass filtered image was used for the combination in each row, and the same high-pass filtered image in each column. The combination of the bandpass and high-pass filtered output images was achieved by a weighted addition of the filtered images, with the bandpass filtered images being weighted by a factor of 0.2 and the high-pass filtered images being weighted by a factor of 0.8 were weighted. In this example, the edge length of the individual images should correspond to 580 arc minutes. Figure 7 shows tables with calculated similarities of the source images B1 to B6 with the combined images shown in Figure 6 when viewed without optical blur. For example, the upper left table of Figure 7 shows the values ​​calculated in this way for the similarity of the combined images shown in Figure 6 with the source image B1. Accordingly, the lower right table of Figure 7 shows the calculated values ​​for the similarity of the combined images shown in Figure 6 with the source image B6. For example, as can be seen from the upper left table of Figure 7, the calculated similarity of image “12” (row 1, column 2) with the source image B1 is 0.013, while the calculated similarity of image “21” (row 2, column 1) with the source image B1 is 0.098.As expected, the calculated similarity of image "11" (row 1, column 1) with the source image B1 is 1,000. The values ​​in the other tables in Figure 7 should be read accordingly. The highest values ​​in a row or column are shown in italics or underlined in the tables. The calculation of the similarities given in the tables in Figure 7 was carried out using a correlation-based similarity function. The similarity function was applied to an original image and a combined filtered image in order to quantitatively determine the similarity of these two images. An embodiment of the definition or determination of a similarity using a similarity function is described below.

[0130] On the left side of Figure 8, for the purposes of the examples described below, the Zernike coefficients of an exemplary wavefront error of the at least one eye of a test subject. Furthermore, on the right-hand side of Figure 8 at the top, a 1024 x 1024 pixel image of the point spread function of the uncorrected eye with an entrance pupil radius of 2.52 mm for a wavelength of 550 nm is shown. An enlargement of this image is shown immediately below the image. The intensity of the point spread function has been normalized for better representation. For simulation purposes, all orders with n ≤ 1 can also be set to zero. In the example shown, the refractive error of the eye is M = -3.94 dpt, Jo = -0.03 dpt and J45 = -0.27 dpt. The point spread function is shown from the perspective of the person with the aforementioned wavefront errors and corresponds to the visual impression of a point light source. Figure 9 shows superimposed filtered images with different optically induced blurs.In each of the five columns of Figure 9, two superimposed filtered images are shown, each viewed at a certain optically induced blur P. i(with i = 1, 2, 3, 4, 5). The M component is as indicated under the respective images. The examples shown are based on an eye with the Zernike coefficients indicated in Figure 8. This eye, with an exemplary pupil radius of 2.52 mm, has a refractive error of M = -3.94 dpt, J0 = -0.03 dpt, and J45 = -0.27 dpt. The higher-order aberrations are as indicated in the table in Figure 8. If this refractive error is corrected with an eye correction having an effect of Mi = -6 dpt, M2 = -5 dpt, M3 = -4 dpt, M4 = -3 dpt and Ms = -2 dpt with astigmatic components of Jo = 0 dpt and J45 = -0.25 dpt at a corneal vertex distance of -3.5 mm (position of the entrance pupil), the optically induced blurs P1 to Ps result with the M values ​​given in Figure 9.One of the two superimposed filtered images in Figure 9 is based on the original image B1 ("elephant"), and the other of the two superimposed filtered images in Figure 9 is based on the original image B2 ("cat"). The images in the upper row of Figure 9 are each an overlay of the images F1(B1) and F2(B2) from Figure 5, while the images in the lower row of Figure 9 are each an overlay of the images F2(B1) and F1(B2) from Figure 5. As can be seen from Figure 9, depending on the filters applied to the superimposed filtered image and the optical blur with which the superimposed image is viewed, the superimposed image is more similar to the original image B1 ("elephant") or to the original image B2 ("cat"). All images based on the “new symbols” (cat, elephant, etc.) and calculated similarity values ​​refer to a view in which the side length of the images corresponds to 580 arc minutes of visual angle.

[0131] Figure 10, similar to Figure 7, shows tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6. Unlike the tables in Figure 7, however, the similarity values ​​contained in the tables in Figure 10 refer to an observation with the optical blur P1 (M = -1.93 dpt, J0 = -0.03 dpt, J45 = -0.02 dpt). The highest values ​​in a row or column are also shown in italics or underlined in the tables in Figure 10.

[0132] Figure 11 shows, similar to Figure 7, tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6. Unlike in the tables of Figure 7, the similarities in the tables of Figure

[0133] However, the similarity values ​​contained in Figure 11 are based on an observation with the optical blur P2 (M = -0.93 dpt, Jo = -0.03 dpt, J45 = -0.02 dpt). The highest values ​​in a row or column are also shown in italics or underlined in the tables in Figure 11.

[0134] Figure 12 shows, similar to Figure 7, tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6. Unlike in the tables of Figure 7, the similarities in the tables of Figure

[0135] However, the similarity values ​​contained in Figure 12 are based on an observation with optical blur P3 (M = 0.06 dpt, Jo = -0.03 dpt, J45 = -0.02 dpt). The highest values ​​in a row or column are also shown in italics or underlined in the tables in Figure 12.

[0136] Figure 13 shows, similar to Figure 7, tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6. Unlike in the tables of Figure 7, the similarities in the tables of Figure

[0137] However, the similarity values ​​contained in Figure 13 are based on an observation with optical blur P4 (M = 1.06 dpt, Jo = -0.03 dpt, J45 = -0.02 dpt). The highest values ​​in a row or column are also shown in italics or underlined in the tables in Figure 13.

[0138] Figure 14 shows, similar to Figure 7, tables with calculated similarities of the original images B1 to B6 with the combined images shown in Figure 6. Unlike in the tables of Figure 7, the similarities in the tables of Figure

[0139] However, the similarity values ​​contained in Figure 14 are based on an observation with the optical blur P5 (M = 2.06 dpt, Jo = -0.03 dpt, J45 = -0.02 dpt). The highest values ​​in a row or column are also shown in italics or underlined in the tables in Figure 14. Figure 15 shows a graphical representation of the similarities from the previous tables in Figures 7 and 10 to 14. In each sub-image of Figure 15, the similarity of the respective source image specified in the title of the sub-image (“moon”, “umbrella”, “sun”, “elephant”, “cat”, “human”) is shown with the combined filtered images viewed through different optically induced blurs (only the dependence on the spherical equivalent is shown). Triangles and circles represent combined filtered images, for the generation of which the respective source image was high-pass or band-pass filtered. Crosses represent combined filtered images that were not generated using the original image.The similarity of the original image decreases rapidly with a spherical equivalent deviating from 0, or less rapidly if the respective original image was high-pass or band-pass filtered to generate the filtered combined image. Example of defining or determining similarity.

[0140] As already mentioned above, the similarity can be specified, in particular, by a similarity function S(B1,B2) of two images B1 and B2. Such a similarity function S(B1,B2) can be determined, for example, from the ratio of the mean decay of the cross-correlation function of the images (after background subtraction), DCF(B1,B2), and the mean of the mean decay of the autocorrelation function of the respective images (also after background subtraction), DCF(B1,B1) and DCF(B2,B2). The similarity function S(B1,B2) can thus be defined, for example, as follows:

[0141] Note that in the formula given above, B1 and B2 refer to images in general. B1 and B2 do not have to be the original images, but can be any image, including filtered images. The mean decay of the cross-correlation function, DCF(B1,B2), or the autocorrelation functions DCF(B1,B1) and DCF(B2,B2), for a shift of one pixel up, down, right, and left can be calculated as follows: where x0and y0the two-dimensional pixel of the maximum of the correlation function CF(B i , Bj) describe which are The summation takes place over all pixels of the image with the positions (m, n). Here, B denotes i [m,n] the intensity of the i-th image at position (m, n) and B t the mean intensity over all pixels of the i-th image. Instead of the maximum x0 and y0 of the correlation function CF(P i , P j) the pixel without displacement can be chosen instead, ie x0= 0 and y0= 0. M and N here refer to the size of the images in pixels (M x N pixels).

[0142] Example of defining or quantifying the similarity of two images based on the classification probability of a classifier (e.g., neural network or other machine learning model):

[0143] A classifier for quantifying the similarity between two images could, for example, be a deep convolutional neural network. This network has been pre-trained using existing image databases (e.g., ImageNet), and whose last layers have been trained using a data set containing the source images with their correct labels in order to correctly assign the source images. In the embodiment shown in Figures 1a to 3, this would be, for example, the original image of Kermit with the label “Kermit” and the original image of Ms. Piggy with the label “Ms. Piggy,” as well as typically other source images. The artificial neural network is trained with the aim of predicting the label from the source images, in other words of solving a typical classification task. In the embodiment shown in Figures 1a to 3, the different possible classes, among which “Kermit” and “Ms.Piggy" are present in a so-called "one-hot" representation, e.g., "Kermit" = (1, 0, 0, ...) and "Ms. Piggy" = (0, 1, 0, 0, ...). The vector output by the artificial neural network can therefore be interpreted as a vector of probabilities for classifying the image fed into the input layer into the candidate classes (as a "one-hot" representation) with the help of the "loss function" used during training (typically the "categorical cross-entropy").

[0144] If, for example, the original image of Kermit is fed into the network, then an output vector (0.96, 0.05, 0.06, 0.01 , ... ) can be output. If the original image of Ms. Piggy is fed into the network, then the output vector (0.03, 0.92, 0.08, 0.02, ... ) results. The numerical values ​​used here are only examples. If any other image is fed into the network, the network outputs an output vector whose k-th element can be understood as the similarity of the k-th original image with the filtered image, or the recognizability of the k-th original image in the filtered image. Any other image can be a filtered image assigned to a first optically induced blur, in which in addition to filtering with a filter function the viewing effect was simulated by a second optically induced blur.The k-th element of the output vector can be considered as the detectability of the k-th output image in the filtered image associated with the first optically induced blur when viewed through the second optically induced blur.

Claims

Patent claims 1. Visual object (100) for determining one or more visual properties of a subject, comprising at least two different filtered images combined into a test image, wherein each of the filtered images represents one of at least two specified different source images and is provided with a filter specific for a specified optically induced blur.

2. Visual object (100) according to claim 1, wherein each of the filtered images is provided with a filter specific for a predetermined optically induced blur such that, when viewed with a first predetermined optically induced blur, a first of the at least two output images is more clearly recognizable than any of the remaining output images that differ from the first output image, and when viewed with a second predetermined optically induced blur, a second of the at least two output images is more clearly recognizable than any of the remaining output images that differ from the second output image.

3. Visual object (100) according to claim 1 or 2, wherein each of the filtered images is provided with a filter specific for a predetermined optically induced blur such that the filtered image, when viewed with the predetermined specific optically induced blur, exhibits a predetermined similarity and / or a predetermined recognizability with its represented original image, in particular a predetermined minimum similarity and / or a predetermined minimum recognizability.

4. Visual object (100) according to any of the preceding claims, wherein the at least two filtered images comprise a first filtered image F1(B1) and a second filtered image F2(B2), wherein the first filtered image F1(B1) represents a first output image B1 filtered with a first filter Fi, wherein the second filtered image F2(B2) represents a second output image B2 filtered with a second filter F2, which differs from the first output image B1, and wherein in particular a similarity S 111 of the first filtered image F1(B1) with its original image B1 when viewed with an initial blur P1 is greater than a similarity S 212 of the second filtered image F2(B2) considered during the first blurring P1 with its original image B2.

5. Visual object (100) according to claim 4, wherein: S 111 > S 121 ; and / or S 222 > S 121 ; and / or S 222 > S 212 ; where S 111 the similarity of the first filtered image F1(B1) considered at the first blurring P1 with its original image B1, where S 222the similarity of the second filtered image F2(B2) considered under a second blurring P2 with its original image B2, where S 121 the similarity of the first filtered image F1(B1) considered in the second blurring P2 with its original image B1, and where S 212 the similarity of the second filtered image F2(B2) considered during the first blurring P1 with its original image B2.

6. Visual object (100) according to claim 4 or 5, wherein furthermore: S 112 < S 111 and S 112 < S 222 ; and / or S122 < S 111 and S 122 < S 222 ; where S 111 the similarity of the first filtered image F1(B1) considered at the first blurring P1 with its original image B1, where S 222 the similarity of the second filtered image F2(B2) considered under a second blurring P2 with its original image B2, where S 112the similarity of the first filtered image F1(B1) considered during the first blurring P1 with the original image B2, and where S122 is the similarity of the first filtered image F1(B1) considered during the second blurring P2 with the original image B2.

7. Visual object (100) according to one of the preceding claims, wherein the at least two filtered images are arranged side by side and / or overlapping, in particular at least partially superimposed.

8. Visual object (100) according to any of the preceding claims, wherein the filters are adapted to average higher-order aberrations of one eye; or wherein the filters are adapted to individual higher-order aberrations of at least one eye of the subject.

9. Methods for determining one or more visual characteristics of a subject, comprising: Providing a visual object (100) according to one of the preceding claims, wherein providing the visual object particularly comprises displaying the visual object, and / or wherein in particular the visual object is presented to the subject.

10. The method of claim 9, wherein providing a visual object (100) comprises: Providing at least two source images (B1, B2); Generating at least two filtered recognition images by applying filters specific for a given optically induced blur to the at least two source images such that each filtered recognition image is assigned to one of the at least two source images, and that when viewing the filtered recognition images with the given optically induced blur, each filtered recognition image exhibits a given similarity, in particular a given minimal similarity, with its assigned source image.

11. Method according to claim 9 or 10, wherein providing the visual object (100) comprises determining the filter specific for a given optically induced blur for each output image, and wherein determining the filter specific for a given optically induced blur for each output image particularly comprises calculating an optical transfer function associated with the respective optically induced blur.

12. Method according to any one of claims 9 to 11, wherein providing the visual object (100) comprises the following steps: Filtering a multitude N of different source images Bi with 1 ≤ i ≤ N, for a multitude M of optically induced blurs Pj with 1 ≤ j ≤ M, where M and N are each positive integers with M < N, and Selecting those filtered images F j (B i ), whose similarity S ijWhen viewed with an optically induced blur Pi, the initial images Bi exceed a first threshold value, and their similarity S ij to the other initial images B j with j ≠ i when viewed with an optically induced blur P i falls below a second threshold.

13. The method of claim 12, wherein the selection of filtered images represents an optimization over all subsets with M source images, wherein the optimization is preferably performed by maximizing an objective function, and wherein the objective function is defined in particular as follows: Where ö represents the Kronecker symbol and w ijk Weights are denoted by w ijk > 0 and ∑ ijk W ijk > 0.

14. Method according to any one of claims 9 to 13, further comprising: Evaluating a subject's reaction to the visual object (100), whereby the evaluation particularly - determining ametropia of the subject, and / or - determining the subject's sensitivity to optically induced blurring, and / or - includes determining an optimal filter for the test subject.

15. Computer program product comprising computer-readable instructions which, when loaded into a memory of a computer and executed by the computer, cause the computer to perform a method according to any one of claims 9 to 14.

Images