A method for generating and using noise for fine-scale discrimination of visual features

Abstract

The present invention discloses a noise generation and use method for a visual feature fine resolution task, comprising the steps of: S1, obtaining parameters of a visual display device; S2, obtaining parameters of a visual feature image to be resolved; S3, generating a random point noise image and a grating noise image according to the parameters of the visual display device and the parameters of the visual feature image to be resolved; S4, obtaining a noise image according to the random point noise image and the grating noise image; S5, superimposing the noise image and the visual feature image to be resolved to obtain a noise superimposed image, and performing a visual feature fine resolution task. The present invention accurately constructs random noise and grating noise for masking the basic visual features "spatial frequency + orientation + contrast", and can be adjusted within a certain parameter range, can match biological individuals and artificial intelligence systems with different visual feature resolution capabilities, has a wide range of applications, is simple and convenient, has strong practicality, and has broad application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of fine resolution of visual features, and in particular to a method for generating and using noise applied to a task of fine resolution of visual features. Background Art

[0002] Feature resolution and content recognition of visual images are important research topics in vision science and artificial intelligence science. The resolution of visual features in images is the basis and prerequisite for content recognition. Therefore, whether it is a real biological individual or a system designed and manufactured based on artificial intelligence technology with visual image recognition capabilities, it should have the ability to distinguish basic visual features.

[0003] Visual noise can mask specific visual features in visual images and can be used to evaluate the visual feature resolution and image recognition capabilities of biological individuals or artificial intelligence systems. Therefore, designing noise similar to the visual features to be resolved and applying it to the task of fine-resolution visual feature resolution is of great significance for the evaluation of the fine-resolution visual feature resolution capabilities of biological individuals or artificial intelligence systems. However, the existing visual noise design methods still need to be improved in terms of the effectiveness of targeted masking of specific visual features.

[0004] In reality, a method for generating and using noise for the task of fine-tuning visual features is needed, which has a wider scope of application and is simpler and more convenient. It can be applied to the evaluation of the fine-tuning ability of visual features of biological individuals or artificial intelligence systems at all levels. Summary of the invention

[0005] The purpose of the present invention is to provide a noise generation and use method for the task of fine resolution of visual features. The method has a wide range of applications, is simple and convenient, and has strong practicality. It can be used to evaluate the fine resolution ability of visual features of biological individuals or artificial intelligence systems at all levels.

[0006] The object of the present invention can be achieved by the following technical solution: A noise generation and use method for a visual feature fine resolution task, comprising the steps of:

[0007] S1. Obtain the parameters of the visual display device, including the maximum brightness L max , minimum brightness L min and the visual angle DPD occupied by each point in the displayed image;

[0008] S2. Obtain the visual feature image L to be distinguished target (x,y) parameters, including spatial frequency sf target 、DirectionOri target and contrast C target ;

[0009] S3, generating a random point noise image L according to the parameters of the visual display device and the parameters of the visual feature image to be distinguished RandomDot (x,y) and the raster noise image L grating (x,y);

[0010] S4, based on the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y);

[0011] S5, superimpose noise image L noise (x, y) and the visual feature image L to be distinguished target (x,y) Get the noise superposition image L mixed (x,y), perform the task of fine-grained visual feature resolution.

[0012] Furthermore, it also includes:

[0013] S6. Adjust the execution mode of the visual feature fine resolution task according to the parameters of the visual display device.

[0014] Furthermore, the random point noise image L is generated in S3 RandomDot The process of (x,y) includes:

[0015] S31a, based on sf target and C target Determine the parameters of the smallest unit in the random point noise image, including the size of the smallest unit RandomDot and the contrast of the smallest component unit C RandomDot , the formula is:

[0016] Size RandomDot =round(DPD / (sf target ×n sf，RandomDot ))

[0017] C RandomDot =min(n c，RandomDot ×C target ,1)

[0018] Among them, round is the rounding function, min is the minimum value function, DPD is the viewing angle occupied by each point (x, y) in the image when the visual display device displays the image (in "units / degree"), n c，RandomDot and n sf，RandomDot All are coefficients;

[0019] S32a, based on the size of the smallest component unit RandomDot and the contrast of the smallest component unit C RandomDot, construct a random point noise image L RandomDot (x,y), the formula is:

[0020] L RandomDot (x,y)=L mean ×(1+R(Size RandomDot , C RandomDot ))

[0021] Among them, L RandomDot (x, y) represents the brightness of each point (x, y) in the random point noise image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated using a random algorithm; the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly taken from a normal distribution N(0,C RandomDot 2 ).

[0022] Furthermore, the process of generating the grating noise image in S3 includes:

[0023] S31b, based on C target 、sf target and Ori target , determine the contrast C of the grating noise image grating , spatial frequency sf grating Ori grating , the formula is:

[0024] C grating =min(n c，grating ×C target ,1)

[0025] sf grating =sf target ×n sf，grating

[0026] Ori grating =Ori target +Ori θ

[0027] Among them, min is the minimum value function; n c，grating and n sf，grating are coefficients, randomly selected from uniform distributions U([1.0,2.0]) and U([0.67,0.8],[1.25,1.66]) respectively; Ori θ is the additional value of orientation, which is randomly selected from the uniform distribution U([-10,10]) and the unit is circular degree;

[0028] S32b, based on contrast C grating , spatial frequency sf grating Ori grating , construct the random grating noise image L grating,1 (x,y), the formula is:

[0029] L grating,1 (x,y)=L mean ×(0.5+C grating ×sin(2π(sf grating ×DPD×(ycos(Ori grating )–xsin(Ori grating )))+Φ))

[0030] Among them, L grating,1 (x, y) represents the brightness at each point (x, y) in a random grating noise image, L mean is the average brightness, DPD is the viewing angle occupied by each point (x, y) in the grating noise image (in units of “pieces / degree”), and Ф is the random value of the grating phase.

[0031] Furthermore, it also includes:

[0032] S33b, repeat S31b to S32b process n times to obtain n L grating,i (x,y), according to n L grating,i (x,y) Get the grating noise image L grating (x,y), the formula is:

[0033]

[0034] Among them, L grating,i (x, y) is the random grating noise image constructed for the i-th time, 1≤i≤n, i and n are positive integers.

[0035] Furthermore, in S4, based on the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y), the formula is:

[0036] L noise (x,y)=L RandomDot (x,y)+L grating (x,y)

[0037] Among them, L noise (x,y) represents the brightness at each point (x,y) in the noise image.

[0038] Furthermore, the contrast of the smallest component unit in S32a is randomly taken from the normal distribution N(0,C RandomDot 2 ), the minimum number of constituent units shall not be less than 256.

[0039] Furthermore, in S6, the execution mode of the visual feature fine resolution task is adjusted according to the parameters of the visual display device, and the process is as follows:

[0040] S61, based on L noise (x,y) and L target (x, y), get the noise superposition image L of the two mixed (x,y), the formula is:

[0041] L mixed (x,y)=L noise (x,y)+L target (x,y)

[0042] Among them, L noise (x, y) represents the brightness of each point (x, y) in the noise image, L target (x, y) represents the brightness at each point (x, y) in the target image of the visual feature to be distinguished;

[0043] S62, judge L mixed Is there a brightness in (x,y) that exceeds the maximum brightness L of the visual display device? max Or below the minimum brightness L min If the point does not exist, the process goes to S63; if it does exist, the process goes to S64;

[0044] S63. In the task of fine-tuning visual features, the noise-superimposed image L is directly presented using a visual display device. mixed (x,y);

[0045] S64. In the visual feature fine discrimination task, a visual display device is used to alternately present noise superimposed images L mixed (x, y) and the visual feature image L to be distinguished target (x,y).

[0046] Furthermore, in step S64, the visual display device is used to alternately present the noise superimposed image L mixed (x, y) and the visual feature image L to be distinguished target When (x,y), each image is presented continuously for a period of 10-100 milliseconds.

[0047] Furthermore, in step S64, the visual display device is used to alternately present the noise superimposed image L mixed (x, y) and the visual feature image L to be distinguishedtarget (x,y), the noise superimposed image L mixed (x,y) cumulative presentation time and visual feature image to be distinguished L target The ratio of (x,y) presentation time is not less than 1.0.

[0048] Beneficial effects of the present invention:

[0049] 1. The present invention accurately constructs random noise and grid noise for masking the basic visual features of "spatial frequency + orientation + contrast", and can be adjusted within a certain parameter range. It can match biological individuals and artificial intelligence systems with different visual feature resolution capabilities. It has a wide range of applications, a simple and convenient method, strong practicality, and broad application prospects.

[0050] 2. The random point noise image and grid noise image of the present invention use n c，RandomDot and n sf，RandomDot Coefficient, n c，grating and n sf，grating Coefficient, randomly selected from uniform distribution U, Ori θ The values ​​are randomly selected from the uniform distribution U([-10,10]). Therefore, when the visual feature fine-resolution task is completed multiple times in a row, the noise used in the task is not exactly the same. The parameters used by the noise can be randomly selected within a certain range. The difficulty of the task after adding noise is much greater than the difficulty of the task before adding. Therefore, when the visual feature fine-resolution task is completed multiple times in a row, the noise used in the task is not exactly the same. It can be used to evaluate the fine-resolution ability of visual features of biological individuals or artificial intelligence systems at a higher level.

[0051] 3. The method for using the noise image of the present invention may use a method of alternately presenting the noise image and the original visual feature image to be distinguished, in addition to directly superimposing the original visual feature image to be distinguished with the noise image; when L mixed (x, y) has a brightness that exceeds the maximum brightness L of the visual display device max Or below the minimum brightness L min At the point in time, the noise image and the original visual feature image to be distinguished can be presented alternately, with each presentation time being 50 milliseconds. This can achieve the goal of "increasing the difficulty of the resolution task by adding noise" while effectively avoiding the display deviation caused by the theoretical brightness in the image exceeding the display range of the visual display device when the noise image is directly superimposed on the original visual feature image to be distinguished. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 A schematic diagram of the noise generation effect of the visual feature fine resolution task of the present invention;

[0053] Figure 2A schematic diagram of the random noise generation effect of the visual feature fine resolution task of the present invention;

[0054] Figure 3 It is a schematic diagram of the effect of alternately presenting a noise image and an original visual feature image to be distinguished according to the present invention. DETAILED DESCRIPTION

[0055] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar symbols throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0056] like Figure 1-Figure 3 As shown, the present invention discloses a noise generation and use method for a visual feature fine resolution task, comprising the steps of:

[0057] S1. Obtain the parameters of the visual display device, including the maximum brightness L max , minimum brightness L min and the visual angle DPD occupied by each point in the displayed image;

[0058] S2. Obtain the visual feature image L to be distinguished target (x,y) parameters, including spatial frequency sf target 、DirectionOri target and contrast C target ;

[0059] S3, generating a random point noise image L according to the parameters of the visual display device and the parameters of the visual feature image to be distinguished RandomDot (x,y) and the raster noise image L grating (x,y);

[0060] S4, based on the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y);

[0061] S5, superimpose noise image L noise (x, y) and the visual feature image L to be distinguished target (x,y) Get the noise superposition image L mixed (x,y), perform the task of fine-grained visual feature resolution.

[0062] S6. Adjust the execution mode of the visual feature fine resolution task according to the parameters of the visual display device.

[0063] The present invention Figure 1The embodiment shown illustrates the method of the present invention:

[0064] The content of the visual feature fine discrimination task is to judge the fine differences of visual features appearing in a specified spatial position and time window. For example, a curve that is close to a straight line appears in a specified spatial position and time window, such as Figure 1 (a) may be slightly curved to the right, such as Figure 1 As shown in (b), it may also be slightly curved to the left. A biological individual or artificial intelligence system that completes the task of finely distinguishing visual features needs to analyze the image and determine the image.

[0065] In the absence of additional noise, the image difference is more obvious, such as Figure 1 (a) and Figure 1 As shown in (b), the difficulty of completing the task of fine-tuning visual features is relatively low; in order to increase the difficulty of the task and make it closer to the actual application scenario, Figure 1 (a) with Figure 1 (b) Add random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y).

[0066] Generate random point noise image L RandomDot The process of (x,y) includes:

[0067] Based on sf target and C target Determine the parameters of the smallest component unit in the random point noise image, such as Figure 1 (a) with Figure 1 (b) Image, spatial frequency sf target It can be determined as 12 cycles / degree, contrast C target It can be determined as 0.60 to determine the minimum component unit parameter in the random point noise image.

[0068] Minimum unit parameters, including the size of the minimum unit RandomDot and the contrast of the smallest component unit C RandomDot , the formula is:

[0069] Size RandomDot =round(DPD / (sf target ×n sf，RandomDot ))

[0070] C RandomDot =min(n c，RandomDot ×C target ,1)

[0071] Among them, round is a rounding function, min is a minimum value function, DPD is the viewing angle occupied by each point (x, y) in the image when the visual display device displays the image (in units of "points / degrees"), and the value of this embodiment can be 96, n c，RandomDot and n sf，RandomDot are coefficients, and can be randomly selected from uniform distributions U[1.0,10.0] and U[3.80,5.80], respectively. In this embodiment, n c，RandomDot The value is 1.0, n sf，RandomDot The value is 3.80, Size RandomDot The value is 2, C RandomDot The value is 0.20.

[0072] Based on the size of the smallest unit RandomDot and the contrast of the smallest component unit C RandomDot , construct a random point noise image L RandomDot (x,y), the formula is:

[0073] L RandomDot (x,y)=L mean ×(1+R(Size RandomDot , C RandomDot ))

[0074] Among them, L RandomDot (x, y) represents the brightness of each point (x, y) in the random point noise image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated using a random algorithm; the image is composed of multiple square minimum components, and the length and width of the minimum component are both Size RandomDott , the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly taken from a normal distribution N(0,C RandomDot 2 ).

[0075] The number of the smallest component units of the random point noise image is not less than 100. When the contrast of the smallest component unit is randomly taken from the normal distribution N(0,C RandomDot 2 ), the minimum number of constituent units shall not be less than 256.

[0076] In this embodiment, the effect of actually generating a random point noise image and superimposing the original visual feature image to be distinguished is as follows: Figure 1 (c) and Figure 1 As shown in (d), it can be seen that after superimposing the random point noise image, the resolution of the original visual feature image to be resolved becomes more difficult.

[0077] Furthermore, because the identification of image details is also closely related to the image orientation (centered on "vertical", slightly tilted or bent to both sides), a grating noise image containing orientation information can also be generated. The process includes:

[0078] C-based target 、sf target and Ori target , determine the contrast C of the grating noise image grating , spatial frequency sf grating Ori grating , the formula is:

[0079] C grating =min(n c，grating ×C target ,1)

[0080] sf grating =sf target ×n sf，grating

[0081] Ori grating =Ori target +Ori θ

[0082] Among them, min is the minimum value function; n c，grating and n sf，grating are coefficients, randomly selected from uniform distributions U([1.0,2.0]) and U([0.67,0.8],[1.25,1.66]) respectively; Ori θ is the additional value of orientation, which is randomly selected from the uniform distribution U([-10,10]) and the unit is circular degree;

[0083] In this embodiment, n c，grating The value is 1.0, n sf，grating The value is 1.33, C grating The value is 0.20, sf grating The value is 16 cycles / degree, Ori θ is 0 degrees.

[0084] Based on contrast C grating , spatial frequency sf grating Ori grating , construct the random grating noise image L grating,1 (x,y), the formula is:

[0085] L grating,1 (x,y)=L mean ×(0.5+C grating ×sin(2π(sfgrating ×DPD×(ycos(Ori grating )–xsin(Ori grating )))+Φ))

[0086] Among them, L grating,1 (x, y) represents the brightness at each point (x, y) in a random grating noise image, L mean is the average brightness, DPD is the viewing angle occupied by each point (x, y) in the grating noise image (in units of “pieces / degree”), and Ф is the random value of the grating phase.

[0087] Construct random grating noise image L grating,i (x,y) process n times, get n L grating,i (x,y), according to n L grating,i (x,y) Get the grating noise image L grating (x,y), the formula is:

[0088]

[0089] Among them, L grating,i (x, y) is the random grating noise image constructed for the i-th time, 1≤i≤n, i and n are positive integers.

[0090] In the embodiment, the effect after the original visual feature image to be distinguished is superimposed on the grating noise image is as follows: Figure 1 (e) with Figure 1 As shown in (f), it can be found that after superimposing the slightly tilted grating, the original visual feature image to be distinguished seems to be a little distorted. This is actually because the azimuth information carried by the grating interferes with the azimuth discrimination of the original visual feature image to be distinguished. However, due to the difference in spatial frequency information in the horizontal direction between the grating and the original visual feature image to be distinguished, the recognition of the image is not seriously disturbed.

[0091] According to the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y), the formula is:

[0092] L noise (x,y)=L RandomDot (x,y)+L grating (x,y)

[0093] Among them, L noise (x,y) represents the brightness at each point (x,y) in the noise image.

[0094] Based on L noise (x,y) and Ltarget (x, y), get the noise superposition image L of the two mixed (x,y), the formula is:

[0095] L mixed (x,y)=L noise (x,y)+L target (x,y)

[0096] Among them, L noise (x, y) represents the brightness of each point (x, y) in the noise image, L target (x, y) represents the brightness at each point (x, y) in the target image of the visual feature to be distinguished.

[0097] According to the parameters of the visual display device, the execution mode of the visual feature fine resolution task is adjusted. mixed There is no brightness in (x, y) that exceeds the maximum brightness L of the visual display device max Or below the minimum brightness L min In the task of fine-resolution visual feature resolution, the visual display device is used to directly present the noise superposition image L mixed (x,y).

[0098] When L mixed (x, y) has a brightness that exceeds the maximum brightness L of the visual display device max Or below the minimum brightness L min In the task of fine-tuning visual features, a visual display device is used to alternately present noise-superimposed images L mixed (x, y) and the visual feature image L to be distinguished target (x,y).

[0099] Using a visual display device to alternately present noise superimposed images L mixed (x, y) and the visual feature image L to be distinguished target (x, y), each noise image presented is generated independently according to the parameters; the time range for each image to be presented continuously is 10-100 milliseconds; the noise superposition image L mixed (x,y) cumulative presentation time and visual feature image to be distinguished L target The ratio of (x,y) presentation time is not less than 1.0.

[0100] In addition to directly superimposing the original visual feature image to be distinguished with the noise image, the method of alternately presenting the noise image and the original visual feature image to be distinguished can also be used to use the method of alternately presenting the noise image and the original visual feature image to be distinguished; when an image with specific visual features appears in the specified spatial position and time window, for example, when L mixed (x, y) has a brightness that exceeds the maximum brightness L of the visual display device maxOr below the minimum brightness L min The noise image and the original visual feature image to be distinguished can be presented alternately, each presentation time is 50 milliseconds, and each is presented twice, such as Figure 3 As shown, while achieving the goal of "increasing the difficulty of the resolution task by adding noise", it can effectively avoid the display deviation that occurs when the theoretical brightness in the image exceeds the display range of the visual display device when the noise image is directly superimposed on the original visual feature image to be resolved.

[0101] In this embodiment, the final noise image is superimposed on the original visual feature image to be distinguished, and the effect is as follows: Figure 1 (g) and Figure 1 As shown in (h); after superimposing the noise image, the content of the discrimination task changes from the original “distinguishing Figure 1 (a) with Figure 1 (b)”, adjusted to “distinguish Figure 1 (g) with Figure 1 (h)”; because the random point noise image in the noise provides spatial frequency and contrast information close to the original visual feature image to be distinguished, and the grating noise image in the noise provides orientation information close to the original visual feature image to be distinguished, the discrimination of the original visual feature image to be distinguished will be significantly disturbed after the noise is superimposed.

[0102] In addition, the parameters used by the noise can be randomly selected within a certain range, so when completing the visual feature fine-resolution task multiple times in a row, the noise used in the task is not exactly the same. θ Randomly select different values ​​from the uniform distribution U([-10,10]), so there are slight differences in the orientation of the grating in the noise, such as Figure 2 (a) with Figure 2 (b) Figure 2 (c) with Figure 2 (d) Figure 2 (e) with Figure 2 (f) Using three different Ori θ ; On this basis, the discrimination experience related to background or noise is difficult to be effectively shared between different tasks. Therefore, the difficulty of the task after adding noise is much greater than the difficulty of the task before adding it, and can be used to evaluate the fine discrimination ability of visual features of higher-level biological individuals or artificial intelligence systems.

[0103] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

[0104] It should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0105] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0106] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral one; it can be a mechanical connection, an electrical connection, or communication with each other; it can be a direct connection, or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0107] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

Claims

1. A method for generating and using noise for fine-resolution visual feature resolution tasks, characterized in that: Includes steps: S1. Obtain the parameters of the visual display device, including the maximum brightness L max , minimum brightness L min and the visual angle DPD occupied by each point in the displayed image; S2. Obtain the visual feature image L to be distinguished target (x,y) parameters, including spatial frequency sf target 、Ori target and contrast C target ; S3, generating a random point noise image L according to the parameters of the visual display device and the parameters of the visual feature image to be distinguished RandomDot (x,y) and the raster noise image L grating (x,y); S4, based on the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y); S5, superimpose noise image L noise (x, y) and the visual feature image L to be distinguished target (x,y) Get the noise superposition image L mixed (x,y), performs the task of fine-grained visual feature resolution; The random point noise image L is generated in S3 RandomDot The process of (x,y) includes: S31a, based on sf target and C target Determine the parameters of the smallest unit in the random point noise image, including the size of the smallest unit RandomDot and the contrast of the smallest component unit C RandomDot , the formula is: Size RandomDot =round(DPD / (sf target ×n sf，RandomDot )) C RandomDot =min(n c，RandomDot ×C target ,1) Among them, round is the rounding function, min is the minimum value function, DPD is the viewing angle occupied by each point (x, y) in the image when the visual display device displays the image (in "points / degree"), n c，RandomDot and n sf，RandomDot All are coefficients; S32a, based on the size of the smallest component unit RandomDot and the contrast of the smallest component unit C RandomDot , construct a random point noise image L RandomDot (x,y), the formula is: L RandomDot (x,y)=L mean ×(1+R(Size RandomDot ,C RandomDot )) Among them, L RandomDot (x, y) represents the brightness of each point (x, y) in the random point noise image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated using a random algorithm; the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly taken from a normal distribution N(0,C RandomDot2 ); Where n c，RandomDot and n sf，RandomDot are coefficients, randomly selected from uniform distributions U[1.0,10.0] and U[3.80,5.80] respectively; The process of generating the grating noise image in S3 includes: S31b, based on C target 、sf target and Ori target , determine the contrast C of the grating noise image grating , spatial frequency sf grating Ori grating , the formula is: C grating =min(n c，grating ×C target ,1) sf grating =sf target ×n sf，grating or grating =Time target +Times θ Among them, min is the minimum value function; n c，grating and n sf，grating are coefficients, randomly selected from uniform distributions U([1.0,2.0]) and U([0.67,0.8],[1.25,1.66]) respectively; Ori θ is the additional value of orientation, which is randomly selected from the uniform distribution U([-10,10]) and the unit is circular degree; S32b, based on contrast C grating , spatial frequency sf grating Ori grating , construct the random grating noise image L grating,1 (x,y), the formula is: L grating,1 (x,y)=L mean ×(0.5+C grating × sin(2π(sf grating xDPDx(ycos(Ori grating )–xsin(Ori grating )))+Φ)) Among them, L grating,1 (x, y) represents the brightness at each point (x, y) in a random grating noise image, L mean is the average brightness, DPD is the viewing angle occupied by each point (x, y) in the grating noise image (unit: "per point / degree"), and Ф is the random value of the grating phase; Also includes: S33b, repeat S31b to S32b process n times to obtain n L grating,i (x,y), according to n L grating,i (x,y) Get the grating noise image L grating (x,y), the formula is: Among them, L grating,i (x, y) is the random grating noise image constructed for the i-th time, 1≤i≤n, i and n are positive integers.

2. The noise generation and use method for fine-resolution visual feature resolution task according to claim 1, characterized in that: Also includes: S6. Adjust the execution mode of the visual feature fine resolution task according to the parameters of the visual display device.

3. The noise generation and use method for fine-resolution visual feature resolution task according to claim 2, characterized in that: In S4, based on the random point noise image L RandomDot (x,y) and the raster noise image L grating (x,y) Get the noise image L noise (x,y), the formula is: L noise (x,y)=L RandomDot (x,y)+L grating (x,y) Among them, L noise (x,y) represents the brightness at each point (x,y) in the noise image.

4. The method for generating and using noise for fine-resolution visual feature resolution according to claim 3, characterized in that: The contrast of the smallest component unit in S32a is randomly taken from the normal distribution N(0,C RandomDot2 ), the minimum number of constituent units shall not be less than 256.

5. The noise generation and use method for fine-resolution visual feature resolution task according to claim 4, characterized in that: In S6, the execution mode of the visual feature fine resolution task is adjusted according to the parameters of the visual display device, and the process is as follows: S61, based on L noise (x,y) and L target (x, y), get the noise superposition image L of the two mixed (x,y), the formula is: L mixed (x,y)=L noise (x,y)+L target (x,y) Among them, L noise (x, y) represents the brightness of each point (x, y) in the noise image, L target (x, y) represents the brightness at each point (x, y) in the target image of the visual feature to be distinguished; S62, judge L mixed Is there a brightness in (x,y) that exceeds the maximum brightness L of the visual display device? max Or below the minimum brightness L min If the point does not exist, the process goes to S63; if it does exist, the process goes to S64; S63. In the task of fine-tuning visual features, the noise-superimposed image L is directly presented using a visual display device. mixed (x,y); S64. In the visual feature fine discrimination task, a visual display device is used to alternately present noise superimposed images L mixed (x, y) and the visual feature image L to be distinguished target (x,y).

6. A noise generation and use method for a visual feature fine resolution task according to claim 5, characterized in that: In step S64, the visual display device is used to alternately present the noise superimposed image L mixed (x, y) and the visual feature image L to be distinguished target When (x,y), each image is presented continuously for a period of 10-100 milliseconds.

7. The noise generation and use method for fine-resolution visual feature resolution task according to claim 6, characterized in that: In step S64, the visual display device is used to alternately present the noise superimposed image L mixed (x, y) and the visual feature image L to be distinguished target (x,y), the noise superimposed image L mixed (x,y) cumulative presentation time and visual feature image to be distinguished L target The ratio of (x,y) presentation time is not less than 1.0.

Images