A noise generation method for visual feature detection tasks

Abstract

The present invention discloses a noise generation method applied to a visual feature detection task, comprising the following steps: S1, obtaining relevant parameters of a visual display device used in the visual feature detection task, including the maximum brightness Lmax and the minimum brightness Lmin that the device can reach, and the DPD occupied by each point in the image when the display device displays the image; S2, obtaining relevant parameters of the visual feature to be detected in the detection task, including the spatial frequency sftarget of the visual feature to be detected, the orientation Oritarget of the visual feature to be detected, and the contrast Ctarget of the visual feature to be detected; S3, determining the components of the noise based on the information obtained in step S2; S4, generating a noise image based on the information obtained in steps S1-S3. The present invention can be used to construct noise that masks specific visual features in a targeted manner, and can match biological individuals and artificial intelligence systems with different visual feature detection capabilities within a certain range through parameter adjustment.

Description

Technical Field

[0001] The present invention relates to the technical field of visual image detection, and in particular to a noise generation method applied to visual feature detection tasks. Background Art

[0002] The feature detection capability of visual images is an important indicator for evaluating the level of visual function of biological individuals, and is also an important research direction for artificial intelligence systems in the field of visual image processing. Compared with feature detection tasks in noise-free environments, feature detection tasks in noisy or even complex noise environments are more difficult, closer to actual application scenarios, and their results also have broader application prospects. Therefore, it is of great significance to design noise similar to the visual features to be detected, apply it to visual feature detection tasks, and evaluate the visual feature detection capabilities of biological individuals or artificial intelligence systems. However, the visual noise that is currently widely used focuses more on universality, and still has obvious shortcomings in the effect of targeted masking of specific visual features. Summary of the invention

[0003] In order to solve the existing problems, the present invention provides a noise generation method applied to visual feature detection tasks. The specific scheme is as follows:

[0004] A noise generation method for visual feature detection tasks comprises the following steps:

[0005] S1, obtaining relevant parameters of the visual display device used in the visual feature detection task, including the maximum brightness Lmax and minimum brightness Lmin that the device can achieve, and the DPD occupied by each point in the image when the display device displays the image;

[0006] S2, obtain the relevant parameters of the visual features to be detected in the detection task, including the spatial frequency sf of the visual features to be detected target , the orientation of the visual feature to be detected target , the contrast of the visual feature to be detected C target ;

[0007] S3, determining the components of the noise based on the information obtained in step S2;

[0008] S4, generating a noise image based on the information obtained in steps S1-S3.

[0009] Preferably, the method for determining the noise components in step S3 is: if Ori target If missing, it is determined that the noise only contains random dot images; otherwise, it is determined that the noise contains two sub-components: random dot images and raster images.

[0010] Preferably, the specific steps of generating the noise image in step S4 include:

[0011] S41, generating a random dot image of noise;

[0012] S42, if the noise also includes a raster component, generating a raster image;

[0013] S43, generating a final noise image.

[0014] Preferably, the number of the minimum constituent units of the random dot image generated in step S41 is not less than 100.

[0015] Preferably, when the contrast of the smallest component unit in the random point image generated in step S41 is randomly taken from the normal distribution N(0, C RandomDot 2 ), the minimum number of constituent units shall not be less than 256.

[0016] Preferably, step S41 specifically includes the following steps:

[0017] S411, based on sf targe t and C target Determine the parameters of the smallest unit in the random point image, including the size of the smallest unit RandomDot , and the contrast of the smallest component unit C RandomDot , the specific calculation formula is:

[0018] Size RandomDot =round(DPD / (sf target ×n sf，RandomDot ))

[0019] C RandomDot =min(n c，RandomDot ×C target , 1)

[0020] 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 display device displays the image, the unit is "point / degree", 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;

[0021] S412, based on the parameters determined in S411, construct a random point image according to the following formula, the specific calculation formula is:

[0022] L RandomDot(x，y) =L mean ×(1+R(Size RandomDot , C RandomDot ))

[0023] Among them, L RandomDot (x, y) represents the brightness of each point (x, y) in the random point image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated by a random algorithm. The image is composed of multiple square minimum components. The length and width of the minimum component are both Size RandomDot , the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly selected from a normal distribution N(0, C RandomDot 2 ).

[0024] Preferably, step S42 specifically includes the following steps:

[0025] S421, based on C target 、sf target and Ori target Determine the contrast C of the grating grating , spatial frequency sf grating Ori grating , the specific calculation formula is:

[0026] C grating =min(n c，grating ×C target , 1)

[0027] sf grating =sf target

[0028] Ori grating =Ori target

[0029] Among them, min is the minimum value function; n c，grating is the coefficient, randomly selected from the uniform distribution U([1.0,3.0]);

[0030] S422, based on the parameters determined in S421, construct a grating image according to the following formula:

[0031] L grating (x,y)=L mean ×(0.5+

[0032] C grating ×sin(2π(sf grating ×DPD×(ycos(Ori grating )–xsin(Ori grating )))+Φ))

[0033] Among them, L grating (x, y) represents the brightness at each point (x, y) in the raster image; L mean is the average brightness; DPD is the viewing angle occupied by each point (x, y) in the image, the unit is "pieces / degree"; Ф is the random value of the grating phase.

[0034] Preferably, the grating image generated in step S42 is the average value of the multiple grating images generated in steps S421 and S422, that is, Where L grating，n (x, y) represents the grating image generated for the nth time according to steps S421 and S422, where n is a positive integer greater than or equal to 1.

[0035] Preferably, the specific calculation formula for generating the final noise image in step S43 is:

[0036] When the noise consists only of random dot images:

[0037] L noise (x, y) = L RandomDot (x, y);

[0038] When the noise contains two subcomponents: random dot image and raster image:

[0039] L noise (x, y) = L RandomDot (x,y)+L grating (x, y);

[0040] Among them, L noise (x, y) represents the brightness at each point (x, y) in the noise image.

[0041] The present invention also discloses a computer system, including a processor and a storage medium, wherein a computer program is stored on the storage medium, and the processor reads and runs the computer program from the storage medium to execute any of the methods described above.

[0042] The beneficial effects of the present invention are:

[0043] The present invention can be used to construct noise that masks specific visual features in a targeted manner, and can match biological individuals and artificial intelligence systems with different visual feature detection capabilities within a certain range through parameter adjustment. The method is simple, convenient, and highly practical, and has broad application prospects in the field of biological individual function assessment and artificial intelligence application. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 This is a diagram illustrating the noise generation effect for visual feature detection tasks proposed in the present invention.

[0046] Figure 2 A diagram illustrating the method of using noise for visual feature detection tasks proposed in the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Generally speaking, the content of the visual inspection task is to determine whether an image with specific visual features appears within a specified spatial position and time window. For example, within a specified spatial position and time window, a Figure 1 (a) There may also be a completely blank content. Figure 1 (b) The biological individual or artificial intelligence system that performs the detection task needs to analyze the image and determine the category of the image.

[0049] In the absence of additional noise, Figure 1 (a) with Figure 1 (b) is more obvious, and the difficulty of the detection task is lower. In order to increase the difficulty of the detection task and make it closer to the actual application scenario, Figure 1 (a) with Figure 1 (b) Noise is added at the same time. The noise generation method applied to the visual feature detection task includes the following steps:

[0050] S1, obtain the relevant parameters of the visual display device used in the visual feature detection task, including the maximum brightness Lmax and minimum brightness Lmin that the device can achieve, and the DPD occupied by each point in the image when the display device displays the image; since DPD is related to the specific display device, it is temporarily set to 96 here.

[0051] S2, obtain the relevant parameters of the visual features to be detected in the detection task, including the spatial frequency sf of the visual features to be detected target , the orientation of the visual feature to be detected target , the contrast of the visual feature to be detected C target .exist Figure 1 (a), the spatial frequency sf of the visual feature to be detected target Can be determined as 2.0 cycles / degree, orientation Ori target Can be determined as 45 degrees, contrast C target It can be determined as 0.50.

[0052] S3, based on the information obtained in step S2, determine the components of the noise; if Ori target If missing, it is determined that the noise only contains random dot images; otherwise, it is determined that the noise contains two sub-components: random dot images and raster images.

[0053] S4, generating a noise image based on the information obtained in steps S1-S3, the specific steps include:

[0054] S41, generating a random dot image of noise; wherein the number of the smallest component units of the generated random dot image is not less than 100. In addition, when the contrast of the smallest component unit in the generated random dot image is randomly taken from the normal distribution N(0, C RandomDot 2 ), the minimum number of constituent units shall not be less than 256.

[0055] Generating a random image of noise involves the following steps:

[0056] S411, based on sf targe t and C target Determine the parameters of the smallest unit in the random point image, including the size of the smallest unit RandomDot , and the contrast of the smallest component unit C RandomDot , the specific calculation formula is:

[0057] Size RandomDot =round(DPD / (sf target ×n sf，RandomDot ))

[0058] C RandomDot =min(n c，RandomDot ×C target , 1)

[0059] 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 display device displays the image, the unit is "point / degree", n c，RandomDot and nsf，RandomDot are coefficients, randomly selected from uniform distributions U[1.0, 10.0] and U[3.80, 5.80], respectively. c，RandomDot The value is 1.0, n sf，RandomDot The value is 4.50, Size RandomDot The value is 10, C RandomDot The value is 0.50.

[0060] S412, based on the parameters determined in S411, construct a random point image according to the following formula, the specific calculation formula is:

[0061] L RandomDot(x，y) =L mean ×(1+R(Size RandomDot , C RandomDot ))

[0062] Among them, L RandomDot (x, y) represents the brightness of each point (x, y) in the random point image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated by a random algorithm. The image is composed of multiple square minimum components. The length and width of the minimum component are both Size RandomDot , the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly selected from a normal distribution N(0, C RandomDot 2 ).

[0063] In an embodiment, the random dot image actually generated is as follows: Figure 1 As shown in (d), the effect of superimposing the random point image on the original image is as follows Figure 1 (c) as shown.

[0064] S42, if the noise also includes a raster component, generating a raster image, specifically comprising the following steps:

[0065] S421, based on C target 、sf target and Ori target Determine the contrast C of the grating grating , spatial frequency sf grating Ori grating , the specific calculation formula is:

[0066] C grating =min(n c，grating ×C target , 1)

[0067] sfgrating =sf target

[0068] Ori grating =Ori target

[0069] Among them, min is the minimum value function; n c，grating is a coefficient, randomly selected from the uniform distribution U([1.0,3.0]). In this embodiment, n c，grating The value is 1.0, C grating The value is 0.50, sf grating The value is 2.0 cycles / degree, Ori grating is 45 degrees.

[0070] S422, based on the parameters determined in S421, construct a grating image according to the following formula:

[0071] L grating (x,y)=L mean ×(0.5+

[0072] C grating ×sin(2π(sf grating ×DPD×(ycos(Ori grating )–xsin(Ori grating )))+Φ))

[0073] Among them, L grating (x, y) represents the brightness at each point (x, y) in the raster image; L mean is the average brightness; DPD is the viewing angle occupied by each point (x, y) in the image, the unit is "pieces / degree"; Ф is the random value of the grating phase.

[0074] In an embodiment, the actual generated raster image is as follows Figure 1 As shown in (f), the effect of superimposing the original image with the grating image is as follows Figure 1 (e) as shown.

[0075] The grating image generated above is the average value of the multiple grating images generated in steps S421 and S422, that is, Where L grating，n (x, y) represents the grating image generated for the nth time according to steps S421 and S422, where n is a positive integer greater than or equal to 1.

[0076] S43, generating the final noise image. The specific calculation formula is:

[0077] When the noise consists only of random dot images:

[0078] L noise(x, y) = L RandomDot (x, y).

[0079] When the noise contains two subcomponents: random dot image and raster image:

[0080] L noise (x, y) = L RandomDot (x,y)+L grating (x, y);

[0081] Among them, L noise (x, y) represents the brightness at each point (x, y) in the noise image.

[0082] In an embodiment, the noise image finally generated is as follows: Figure 1 As shown in (h), the effect of superimposing the random point image on the original image is as follows Figure 1 (g) as shown.

[0083] After superimposing the noise image, the detection task changes from the original “distinguishing Figure 1 (a) with Figure 1 (b)”, adjusted to “distinguish Figure 1 (g) with Figure 1 (h)". Through simple observation, it can be found that the difficulty of the task after adjustment is much greater than that before adjustment, which can complete the evaluation of the detection ability of biological individuals or artificial intelligence systems at a higher level.

[0084] In addition to the method of "directly superimposing the original image and the noise image" described above, the method of using the noise image generated above can also be used to alternately present the noise image and the original image. When an image with specific visual features appears in the specified spatial position and time window, the noise image and the original image are alternately presented, each presentation time is 50 milliseconds, and each is presented twice, such as Figure 2 (a) When no image with specific visual features appears in the specified spatial position and time window, the noise image and the blank image are presented alternately, each presentation time is 50 milliseconds, and each is presented twice, as shown in Figure 2 The above method can achieve the goal of "increasing the difficulty of the detection task by adding noise" while effectively avoiding the display deviation caused by the theoretical brightness of the image exceeding the display range of the visual display device when the noise image is directly superimposed on the original image.

[0085] Specifically, the noise utilization method applied to the visual feature detection task includes the following steps:

[0086] ST1, calculate the noise image L noise (x, y) and the target image L with the visual features to be detected target (x,y) The superimposed image L of the two mixed(x,y); the specific formula is L mixed (x,y)=L noise (x,y)+L target (x,y); where 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 with the visual features to be detected.

[0087] ST2, determine the L obtained in step ST1 mixed Is there a brightness in (x, y) that exceeds the maximum brightness L that the visual display device can display? max Or the brightness is lower than the minimum brightness L that the visual display device can display min If the point does not exist, go to ST3; if it does exist, go to ST4.

[0088] ST3, in the visual feature detection task, directly use the visual display device to present the superimposed image L mixed (x,y).

[0089] ST4, in the visual feature detection task, alternately presenting noise images L mixed (x, y) and the target image L with the visual features to be detected target (x,y).

[0090] When the noise image and the target image with the visual features to be detected are presented alternately, the time range for continuous presentation of each image is 10-100 milliseconds; each image is presented at least twice; the ratio of the cumulative presentation time of the noise image to the cumulative presentation time of the target image is not less than 1.0.

[0091] The present invention also discloses a computer-readable storage medium, on which a computer program is stored. After the computer program is run, any of the above methods is executed.

[0092] The present invention also discloses a computer system, including a processor and a storage medium, wherein a computer program is stored on the storage medium, and the processor reads and runs the computer program from the storage medium to execute any of the methods described above.

[0093] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or a combination of the two. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps are generally described above in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. The technician may implement the described functionality in different ways for each specific application, but such implementation decisions should not be interpreted as resulting in a departure from the scope of the present invention.

[0094] The various illustrative logic blocks, modules, and circuits described in conjunction with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in cooperation with a DSP core, or any other such configuration.

[0095] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software module may reside in a RAM memory, a flash memory, a ROM memory, an EPROM memory, an EEPROM memory, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read and write information from / to the storage medium. In an alternative, a storage medium may be integrated into a processor. The processor and the storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and the storage medium may reside in a user terminal as discrete components.

[0096] In one or more exemplary embodiments, the functions described may be implemented in hardware, software, firmware, or any combination thereof. If implemented as a computer program product in software, each function may be stored on or transmitted by a computer-readable medium as one or more instructions or codes. Computer-readable media include both computer storage media and communication media, including any medium that facilitates the transfer of a computer program from one place to another. Storage media may be any available medium that can be accessed by a computer. As an example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of an instruction or data structure and can be accessed by a computer. Any connection is also properly referred to as a computer-readable medium. For example, if the software is transmitted from a website, a server, or other remote source using a coaxial cable, a fiber optic cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. Disk and disc as used herein include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc, wherein disk often reproduces data magnetically, while disc reproduces data optically with lasers. Combinations of the above should also be included within the scope of computer-readable media.

[0097] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

[0098] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A noise generation method for visual feature detection tasks, characterized in that: The following steps are involved: S1, obtaining relevant parameters of the visual display device used in the visual feature detection task, including the maximum brightness Lmax and minimum brightness Lmin that the device can achieve, and the DPD occupied by each point in the image when the display device displays the image; S2, obtain the relevant parameters of the visual features to be detected in the detection task, including the spatial frequency sf of the visual features to be detected target , the orientation of the visual feature to be detected target , the contrast of the visual feature to be detected C target ; S3, determining the components of the noise based on the information obtained in step S2; S4, generating a noise image based on the information obtained in steps S1-S3; The specific steps of generating the noise image in step S4 include: S41, generating a random dot image of noise; S42, if the noise also includes a raster component, generating a raster image; S43, generating a final noise image; Step S41 specifically includes the following steps: S411, based on sf targe t and C target Determine the parameters of the smallest unit in the random point image, including the size of the smallest unit RandomDot , and the contrast of the smallest component unit C RandomDot , the specific calculation 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 display device displays the image, the unit is "point / degree", 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; S412, based on the parameters determined in S411, construct a random point image according to the following formula, the specific calculation 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 image; L mean is the average brightness; R(Size RandomDot , C RandomDot ) is an image generated by a random algorithm. The image is composed of multiple square minimum components. The length and width of the minimum component are both Size RandomDot , the contrast of the smallest component unit is C RandomDot or –C RandomDot Randomly selected with equal probability, or randomly selected from a normal distribution N(0, C RandomDot 2 ); Step S42 specifically includes the following steps: S421, based on C target 、sf target and Ori target Determine the contrast C of the grating grating , spatial frequency sf grating Ori grating , the specific calculation formula is: C grating =min(n c，grating ×C target ,1) sf grating =sf target or grating =Time target Among them, min is the minimum value function; n c，grating is the coefficient, randomly selected from the uniform distribution U([1.0,3.0]); S422, based on the parameters determined in S421, construct a grating image according to the following formula: L grating (x,y)=L mean ×(0.5+ <h2 style=";text-align:left;direction:ltr">C<h2 style=";text-align:left;direction:ltr"> grating <h2 style=";text-align:left;direction:ltr"> ×sin(2π(sf<h2 style=";text-align:left;direction:ltr"> grating <h2 style=";text-align:left;direction:ltr"> ×DPD×(ycos(Ori<h2 style=";text-align:left;direction:ltr"> grating <h2 style=";text-align:left;direction:ltr"> )-xsin(Ori<h2 style=";text-align:left;direction:ltr"> grating <h2 style=";text-align:left;direction:ltr"> )))+Φ)) Among them, L grating (x, y) represents the brightness at each point (x, y) in the raster image; L mean is the average brightness; DPD is the viewing angle occupied by each point (x, y) in the image, the unit is "pieces / degree"; Ф is the random value of the grating phase; The specific calculation formula for generating the final noise image in step S43 is: When the noise consists only of random dot images: L noise (x,y)=L RandomDot (x,y); When the noise contains two subcomponents: random dot image and raster image: 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.

2. The method according to claim 1, characterized in that The method for determining the noise components in step S3 is: if Ori target If missing, it is determined that the noise only contains random dot images; otherwise, it is determined that the noise contains two sub-components: random dot images and raster images.

3. The method according to claim 2, characterized in that: The number of the minimum constituent units of the random dot image generated in step S41 is not less than 100.

4. The method according to claim 3, characterized in that: When the contrast of the smallest component unit in the random point image generated in step S41 is randomly taken from the normal distribution N(0, C RandomDot 2 ), the minimum number of constituent units shall not be less than 256.

5. The method according to claim 4, characterized in that: The grating image generated in step S42 is the average value of the multiple grating images generated in steps S421 and S422, that is, Wherein Lgrating,n(x,y) represents the grating image generated for the nth time according to steps S421 and S422, and n is a positive integer greater than or equal to 1.

6. A computer system, characterized in that: The method comprises a processor and a storage medium, wherein a computer program is stored in the storage medium, and the processor reads and runs the computer program from the storage medium to execute the method as claimed in any one of claims 1 to 5.

Images