Intelligent saliency feature analysis method and system based on unintentional blindness

Abstract

The invention belongs to the field of unintentional visual blindness analysis, and particularly relates to an unintentional visual blindness-based significance feature intelligent analysis method and system, and the analysis method comprises the following steps: S1, defining an attention task and an attention set: selecting the attention task needing to be completed by an observer, and determining the attention set established by the observer according to the task demand; s2, controlling stimulation features: determining features of the unexpected object, and controlling saliency and similarity of the unexpected object, so that quantitative separation and interaction analysis of the saliency features and the attention set are realized, the analysis precision of an unintentional visual blindness mechanism is improved, a dynamic experiment normal form with high ecological efficiency is constructed, and the method is suitable for popularization and application. And the generalization ability and engineering application value of a research result are enhanced, a perception rate-task performance double-index evaluation system is established, multi-dimensional intelligent diagnosis of attention failure is realized, an iterative intelligent analysis framework is proposed, and cross-scene migration and continuous model optimization are supported.

Description

Technical Field

[0001] This invention relates to the field of unintentional visual blindness analysis technology, and in particular to an intelligent analysis method and system based on the salient features of unintentional visual blindness. Background Technology

[0002] With the convergence of artificial intelligence and cognitive science, saliency detection technology has expanded from traditional image processing to modeling and predicting human visual attention. However, existing saliency models mostly focus on "visible" stimuli, neglecting a crucial phenomenon—inattentional blindness (IB): when an observer's attention is focused on a specific task, even unintended stimuli with high physical saliency may be completely ignored. This phenomenon reveals the complex interplay between "top-down" attentional control and "bottom-up" stimulus-driven processes in the human visual system, and is a core scientific question for understanding the bottlenecks in human visual perception and optimizing the human-computer interaction performance of intelligent systems. Traditional IB research mainly employs a static paradigm (such as the classic "gorilla experiment"), which has low ecological validity, coarse parameter control, and difficulty in quantifying the interaction mechanism between stimulus features and attentional set. While dynamic unintentional visual blindness paradigms (such as the "collision task") introduce motion stimuli, they suffer from the following technical bottlenecks: Dimensional control deficiencies: The correlation between the multidimensional features (color, shape, semantics) of unexpected objects and attention / ignoring items is not systematically decoupled, leading to distorted saliency weight calculations; Cognitive state confusion: The lack of a "full attention" baseline condition makes it impossible to distinguish between the cognitive differences of "not seeing" and "not being seen," resulting in a data contamination rate as high as 15%-30%; Lack of parameter standardization: A reusable standard library of physical properties of stimulus materials (RGB values, collision dynamics parameters) has not been established, leading to low consistency across experiments (Cohen's κ < 0.4); High subjectivity of perception criteria: Reliance on verbal reports and the lack of a quantitative standard for "descriptive sufficiency" cause the incidence of IB to fluctuate by more than 20% in similar experiments. Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of existing technologies by proposing an intelligent analysis method and system based on the salient features of unintentional visual blindness.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: The intelligent analysis method based on the salient features of unintentional visual blindness includes the following steps: S1. Define attentional task and attentional set: Select the attentional task that the observer needs to complete, and clarify the attentional set established by the observer according to the task requirements; S2. Manipulating stimulus features: Identify the features of unexpected objects and control their salience and similarity. S3. Set experimental conditions: Design experimental conditions, pay attention to the characteristics of the relevant and unrelated dimensions of the system operation, ensure the random allocation of experimental conditions, and avoid interference of the experimental order with the results. S4. Plan the experimental procedure: Set up the experimental procedure, including practice experiment, critical experiment and full attention experiment. In each experiment, record whether the observer notices the unexpected object and ask them to describe its characteristics. S5. Data Collection and Analysis: Collect key data, use statistical analysis methods to analyze the differences in detection rates under different experimental conditions, verify the impact of significant features and attentional set on unintentional visual blindness, and analyze the interaction effects. S6. Verification and Optimization: Compare the experimental results with the theoretical assumptions to verify whether they meet expectations. Adjust the experimental design based on the results, optimize the analysis methods, and repeat the experiment to verify the stability of the results. S7. Summarize the patterns: Summarize the main influencing factors of unintentional visual blindness, and propose an intelligent analysis framework based on the salience features of unintentional visual blindness to predict the probability of detecting unexpected objects.

[0005] Preferably, in step S5, the collection of key data includes: Detection rate: Whether the observer notices an unexpected object; Error rate in counting tasks: assessing the observer's performance in attentional tasks.

[0006] Preferably, in step S7, the main influencing factors of unintentional visual blindness are summarized, including: bottom-up stimulus salience, top-down attentional set, and semantic features of the stimulus.

[0007] Preferably, in S4, the practice test familiarizes the observer with the task, the key test introduces an unexpected object and observes whether it is noticed, and the full attention test does not set an attention task and observes whether the unexpected object can be noticed.

[0008] Preferably, in S3, the experimental design only considers the similarity between unexpected objects and attention items, and does not consider whether unexpected objects are similar to or dissimilar to neglected items.

[0009] Preferably, in S3, the experimental procedure follows the dynamic unintentional visual blindness paradigm, combining the dynamic nature of the selective fixation paradigm with the strict control of the static unintentional visual blindness paradigm. Through a series of experiments, the system manipulates the characteristic relationship between attention items and unexpected objects in the relevant and unrelated dimensions of attentional set, explores the relationship between attention and visual perception processing, top-down and bottom-up perception processing, and examines the specific conditions and influencing factors of unintentional visual blindness.

[0010] Preferably, the experimental conditions also include observers and stimulus materials. The observers include 43 observers with normal or corrected visual acuity who participated in the experiment. The observers were divided into groups of 6-8 people to be tested. The data of 2 observers were subsequently removed because the counting data was incomplete. No new objects were found in the full attention test. Based on the characteristic relationship between unexpected objects and attention items, 41 observers were randomly assigned to 4 experimental conditions.

[0011] Preferably, the stimulus material includes: a collision demonstration program written in Microsoft Visual C++ 6.0, presented on a 14-inch monitor of a Founder computer with a screen resolution of 1024. 768 pixels, the background of the display frame is gray, and the size is 582. 472 pixels 18cm 15.12cm, the size of the objects within the frame is 25. 25 pixels 1cm The observer sits approximately 40cm in front of the computer screen, with no fixed head position. The task involves a dynamic, unintentional visual-blindness simulation presented on the computer. The collision demonstration program interface features eight different colored and shaped objects moving randomly at variable rates. These objects move along independent trajectories, periodically hitting the edge of the display frame before smoothly bouncing back. Objects also collide with each other, changing direction and crashing into the frame. Each trial lasts 15 seconds. In each trial, the observer is required to observe a specific shaped object and silently count the number of times it hits the edge. In critical trials, unexpected cyan or red-orange rhomboid objects appear. Each group of objects collides with the frame 30 times in the demonstration interface, ensuring strict control over the number of collisions under different experimental conditions. The RGB values ​​for each color are as follows: Blue RGB: 0,0,255; Green RGB: 0,255,0; Cyan RGB: 112,219,147; Red-orange RGB: 255,36,0.

[0012] Preferably, the questionnaire designed to complement the experiment primarily collects demographic information, records the number of collisions in each trial, whether unexpected objects appeared, and provides a detailed description of new objects. For the item regarding whether observers perceived the appearance of unexpected objects, the following method is used for data analysis: if observers believe they saw a new or different object but their description is insufficient, or they describe irrelevant or unrelated changes, such as changes in the object's movement or color, they are considered unaware of the unexpected object and are therefore considered unperceived; if observers believe they saw a new object and mention at least one correct piece of information in their description of the unexpected object, they are considered to have perceived the unexpected object and are therefore considered perceived. The experimental procedure adopts the previous dynamic unintentional visual blindness paradigm, presenting a collision demonstration procedure in each trial. The experimental procedure is as follows: 8 An object moves within a frame. Each object has two characteristics: shape and color. They are divided into two groups based on their shape. Each observer completes six trials. The instructions vary slightly between trials. During the experiment, the observer is instructed to silently count the number of times the object of the specified color or shape hits the edge of the frame. They are not required to look at other objects within the frame. The first two trials do not include unexpected objects. Unexpected stimuli appear in the last three trials, but the instructions change. In the key trial, after the objects within the frame have moved for 5 seconds, an unexpected stimulus will cross the center of the display from the left side of the screen and disappear on the right side. The entire process lasts for 5 seconds. In the distributed attention experiment, although observers were not explicitly told that new objects would appear, the instructions after each key experiment altered their expectations of unexpected object appearances. Observers recorded the number of times they bumped into the edge on a questionnaire, answered whether they saw a new object appear, and described the new object in detail on the questionnaire. In the full attention experiment, the instructions changed; observers were not required to silently count the number of times any object bumped into the edge, but only to focus on the movement of objects in the display box. Observers could then fully concentrate their attention on the object in the display box. By focusing the observer on all objects within the entire display frame, unexpected objects will be observed. The observer is then asked if they have seen any new objects. The full attention test controls the observer's understanding and adherence to instructions, ensuring that any failure to report a critical stimulus in the critical test is not interpreted as an inability to see the stimulus. If the observer does not discover any new objects in this test, they are considered to be inattentive or have not understood or followed the instructions, and their data is deleted. Therefore, in the full attention test, the observer's awareness rate is 100%, and there is no record of any instances of "hitting the edge," so this data is not included in the experimental data analysis.

[0013] This invention also proposes an intelligent analysis system based on the salient features of unintentional visual blindness, comprising: Define the module to define attentional tasks and attentional set; The stimulus feature manipulation module is used to determine the features of unexpected objects and control the salience and similarity of unexpected objects. The experimental condition setting module is used to design experimental conditions. The system operation pays attention to the characteristics of the relevant and unrelated dimensions of the set, ensuring the random allocation of experimental conditions and avoiding interference of the experimental order with the results. The experimental procedure planning module is used to set up the experimental procedure, including practice experiments, critical experiments, and full attention experiments. In each experiment, it records whether the observer notices an unexpected object and requires the observer to describe its characteristics. The data acquisition and analysis module is used to collect key data, use statistical analysis methods to analyze the differences in detection rates under different experimental conditions, verify the impact of significant features and attentional set on unintentional visual blindness, and analyze the interaction effects. The verification and optimization module is used to compare experimental results with theoretical assumptions, verify whether they meet expectations, adjust the experimental design based on the results, optimize the analysis methods, and repeat the experiment to verify the stability of the results.

[0014] The beneficial effects of the intelligent analysis method and system based on the salient features of unintentional visual blindness in this invention are as follows: This method achieves quantitative separation and interactive analysis of saliency features and attentional set, improving the analytical accuracy of unintentional visual blindness mechanisms: By systematically manipulating the stimulus features of relevant and irrelevant dimensions of attentional set, and combining the dynamic unintentional visual blindness paradigm, this method achieves for the first time the independent variable control and interactive effect quantitative analysis of bottom-up saliency and top-down attentional set, significantly improving the modeling accuracy of the conditions for unintentional visual blindness and providing theoretical support for intelligent systems to predict "visual omissions".

[0015] A dynamic experimental paradigm with high ecological validity is constructed to enhance the generalization ability and engineering application value of research results: A dynamic collision demonstration program is used to simulate real multi-object interaction scenarios. Combining random motion trajectories, edge bounce, and object collision mechanisms, the attention allocation process in complex visual environments is reproduced while strictly controlling stimulus parameters. This paradigm has higher ecological validity than traditional static stimuli, and its conclusions can be directly applied to intelligent scenarios such as autonomous driving early warning systems and anomaly detection in security monitoring, reducing the risk of missed detections due to "unintentional blindness."

[0016] A dual-indicator evaluation system of "awareness rate - task performance" is established to achieve multi-dimensional intelligent diagnosis of attention failure: By simultaneously collecting the awareness rate of unexpected objects and the error rate of counting tasks, this method not only identifies "whether it is seen" but also quantifies "whether attentional resources are adequately allocated." Combined with the quality grading standards of questionnaire descriptions, it can intelligently distinguish between two states: "truly not aware" and "aware but incomplete reporting," providing a real-time diagnostic basis for user status in the development of adaptive attention reminder systems.

[0017] This paper proposes an iterative intelligent analysis framework that supports cross-scenario transfer and continuous model optimization. Through a closed-loop process of "experiment-validation-optimization," this method derives a three-dimensional influencing factor model of "stimulus saliency × attentional set × semantic features," forming a scalable framework for predicting unintentional visual blindness. This framework supports rapid transfer to new scenarios: only the stimulus materials need to be adjusted, and the original experimental logic and data analysis process can be reused to achieve rapid calibration with small samples and continuous model optimization, significantly reducing the deployment cost of intelligent sensing systems.

[0018] This invention achieves quantitative separation and interactive analysis of salient features and attentional set, improves the analytical accuracy of unintentional visual blindness mechanisms, constructs a dynamic experimental paradigm with high ecological validity, enhances the generalization ability and engineering application value of research results, establishes a dual-indicator evaluation system of "awareness rate-task performance", realizes multi-dimensional intelligent diagnosis of attentional failure, proposes an iterative intelligent analysis framework, and supports cross-scenario transfer and continuous model optimization. Attached Figure Description

[0019] Figure 1 This is a flowchart of the intelligent analysis method for salient features based on unintentional visual blindness proposed in this invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Example 1 Reference Figure 1 The intelligent analysis method based on the salient features of unintentional visual blindness includes the following steps: S1. Define attentional task and attentional set: Select the attentional task that the observer needs to complete, and clarify the attentional set established by the observer according to the task requirements; S2. Manipulating stimulus features: Identify the features of unexpected objects and control their salience and similarity. S3. Set experimental conditions: Design experimental conditions, pay attention to the characteristics of the relevant and unrelated dimensions of the system operation, ensure the random allocation of experimental conditions, and avoid interference of the experimental order with the results. S4. Plan the experimental procedure: Set up the experimental procedure, including practice experiment, critical experiment and full attention experiment. In each experiment, record whether the observer notices the unexpected object and ask them to describe its characteristics. S5. Data Collection and Analysis: Collect key data, use statistical analysis methods to analyze the differences in detection rates under different experimental conditions, verify the impact of significant features and attentional set on unintentional visual blindness, and analyze the interaction effects. S6. Verification and Optimization: Compare the experimental results with the theoretical assumptions to verify whether they meet expectations. Adjust the experimental design based on the results, optimize the analysis methods, and repeat the experiment to verify the stability of the results. S7. Summarize the patterns: Summarize the main influencing factors of unintentional visual blindness, and propose an intelligent analysis framework based on the salience features of unintentional visual blindness to predict the probability of detecting unexpected objects.

[0022] In this embodiment, in step S5, key data is collected, including: Detection rate: Whether the observer notices an unexpected object; Error rate in counting tasks: assessing the observer's performance in attentional tasks.

[0023] In this embodiment, in S7, the main influencing factors of unintentional visual blindness are summarized, including: bottom-up stimulus salience, top-down attentional set, and semantic features of stimuli.

[0024] In this embodiment, in S4, the practice test familiarizes the observer with the task, the key test introduces an unexpected object and observes whether it is noticed, and the full attention test does not set an attention task and observes whether the unexpected object can be noticed.

[0025] In this embodiment, in S3, the experimental design only considers the similarity between unexpected objects and attention items, and does not consider whether unexpected objects are similar or dissimilar to ignore items.

[0026] In this embodiment, in S3, the experimental procedure follows the dynamic unintentional visual blindness paradigm. Combining the dynamic nature of the selective fixation paradigm with the strict control of the static unintentional visual blindness paradigm, a series of experiments are conducted to systematically manipulate the characteristic relationship between attention items and unexpected objects in the dimensions related and unrelated to attention set. The relationship between attention and visual perception processing, top-down and bottom-up perception processing is explored, and the specific conditions and influencing factors of unintentional visual blindness are examined.

[0027] In this embodiment, the experimental conditions also include observers and stimulus materials. The observers include 43 observers with normal or corrected normal vision who participated in the experiment. The observers were divided into groups of 6-8 people to be tested. The data of 2 observers were subsequently removed because the counting data was incomplete. No new objects were found in the complete attention test. Based on the characteristic relationship between unexpected objects and attention items, 41 observers were randomly assigned to 4 experimental conditions.

[0028] In this embodiment, the stimulus material includes: a collision demonstration program written in Microsoft Visual C++ 6.0, presented on a 14-inch monitor of a Founder computer with a screen resolution of 1024. 768 pixels, the background of the display frame is gray, and the size is 582. 472 pixels 18cm 15.12cm, the size of the objects within the frame is 25. 25 pixels 1cm The observer sits approximately 40cm in front of the computer screen, with no fixed head position. The task involves a dynamic, unintentional visual-blindness simulation presented on the computer. The collision demonstration program interface features eight different colored and shaped objects moving randomly at variable rates. These objects move along independent trajectories, periodically hitting the edge of the display frame before smoothly bouncing back. Objects also collide with each other, changing direction and crashing into the frame. Each trial lasts 15 seconds. In each trial, the observer is required to observe a specific shaped object and silently count the number of times it hits the edge. In critical trials, unexpected cyan or red-orange rhomboid objects appear. Each group of objects collides with the frame 30 times in the demonstration interface, ensuring strict control over the number of collisions under different experimental conditions. The RGB values ​​for each color are as follows: Blue RGB: 0,0,255; Green RGB: 0,255,0; Cyan RGB: 112,219,147; Red-orange RGB: 255,36,0.

[0029] In this embodiment, the questionnaire designed to complement the experiment primarily collects demographic information, records the number of collisions in each trial, whether unexpected objects appeared, and provides a detailed description of new objects. For the item regarding whether observers perceived the appearance of unexpected objects, the following method was used for data analysis: If an observer believes they saw a new or different object but their description is insufficient, or describes unrelated or irrelevant changes, such as changes in the object's movement or color, they are considered unaware of the unexpected object and are therefore considered unaware. If an observer believes they saw a new object and mentions at least one correct piece of information in their description of the unexpected object, they are considered to have perceived the unexpected object and are therefore considered aware. The experimental procedure adopts the previous dynamic unintentional visual blindness paradigm. Each trial presents a collision demonstration program. The experimental procedure is as follows: 8 objects move within a frame, each object having both shape and color characteristics. They are divided into two groups based on shape characteristics. Each observer completes 6... In this experiment, the instructions varied slightly between different trials. Observers were instructed to silently count the number of times an object of a specified color or shape hit the edge of the screen, without needing to focus on other objects within the frame. The first two trials did not include unexpected objects. Unexpected stimuli appeared in the latter three trials, but the instructions changed. In the critical trial, after the objects in the frame moved for 5 seconds, an unexpected stimulus would cross the center of the screen from the left and disappear on the right, lasting for 5 seconds. After each trial, observers recorded the number of times the object hit the edge on a questionnaire, answered whether they saw a new object appear, and described the new object in detail. In the distributed attention trial, although observers were not explicitly told that a new object would appear, the prompts after the critical trial altered their expectations of the unexpected object's appearance. Observers recorded the number of times the object hit the edge after the trial and were prompted whether they saw a new object appear, then described it in detail on the questionnaire. In the complete attention trial, the instructions changed; observers were not required to silently count the number of times any object hit the edge, but only to focus on the movement of the objects in the display frame. Observers could then fully concentrate their attention on the object. By focusing the observer on all objects within the entire display frame, unexpected objects will be observed. The observer is then asked if they have seen any new objects. The full attention test controls the observer's understanding and adherence to instructions, ensuring that any failure to report a critical stimulus in the critical test is not interpreted as an inability to see the stimulus. If the observer does not discover any new objects in this test, they are considered to be inattentive or have not understood or followed the instructions, and their data is deleted. Therefore, in the full attention test, the observer's awareness rate is 100%, and there is no record of any instances of "hitting the edge," so this data is not included in the experimental data analysis.

[0030] This invention also proposes an intelligent analysis system based on the salient features of unintentional visual blindness, comprising: Define the module to define attentional tasks and attentional set; The stimulus feature manipulation module is used to determine the features of unexpected objects and control the salience and similarity of unexpected objects. The experimental condition setting module is used to design experimental conditions. The system operation pays attention to the characteristics of the relevant and unrelated dimensions of the set, ensuring the random allocation of experimental conditions and avoiding interference of the experimental order with the results. The experimental procedure planning module is used to set up the experimental procedure, including practice experiments, critical experiments, and full attention experiments. In each experiment, it records whether the observer notices an unexpected object and requires the observer to describe its characteristics. The data acquisition and analysis module is used to collect key data, use statistical analysis methods to analyze the differences in detection rates under different experimental conditions, verify the impact of significant features and attentional set on unintentional visual blindness, and analyze the interaction effects. The verification and optimization module is used to compare experimental results with theoretical assumptions, verify whether they meet expectations, adjust the experimental design based on the results, optimize the analysis methods, and repeat the experiment to verify the stability of the results.

[0031] Example 2 The difference between this embodiment and Embodiment 1 is that, in the key experiment, the difference between the time when the unexpected object appears and the time when the observer first reports the awareness is recorded, which is defined as the "awareness delay time".

[0032] By combining eye-tracking data, we can simultaneously analyze the dynamic competition process between salient features (color changes, abnormal motion trajectories) and attentional resource allocation.

[0033] Based on survival analysis modeling, the cumulative probability distribution of detection delay under different experimental conditions is quantified, revealing the critical law of "significance threshold-time window": for example, if a cyan unexpected object is not detected for more than 2.3 seconds when its trajectory deviates from the group's coordination, the subsequent detection probability drops sharply by more than 50%.

[0034] The above approach breaks through the limitations of static awareness rate indicators: traditional methods only record "whether or not it is noticed", while this innovation incorporates the time dimension into the unintentional visual blindness prediction framework for the first time, which can assess in real time when the "risk of not being noticed" reaches the irreversible inflection point.

[0035] Empowering dynamic intervention systems: In intelligent security or autonomous driving, if the system detects that the driver / monitor's perception delay exceeds the critical time window, it can immediately trigger multi-level alarms (such as visual flashing + auditory cues) to transform unintentional blindness into a state of awareness.

[0036] Optimize the saliency calculation model: Provide human time-sensitive parameters for saliency detection algorithms in computer vision, such as requiring AI to forcibly increase the saliency weight of unexpected objects within 1.5 seconds of their appearance (e.g., by enhancing color contrast), thus compensating for the bottleneck of human perception.

[0037] The rest is the same as in Example 1.

[0038] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An intelligent analysis method based on salient features of unintentional blindness, characterized in that, The method comprises the following steps: S1, defining attention tasks and attention set: selecting attention tasks that observers need to complete, and determining the attention set established by the observers according to the task requirements; S2, controlling stimulus characteristics: determining the characteristics of unexpected objects, and controlling the saliency and similarity of unexpected objects; S3, setting experimental conditions: designing experimental conditions, systematically manipulating the characteristics of attention set related dimensions and irrelevant dimensions, ensuring random allocation of experimental conditions, and avoiding interference of experimental order on results; S4, planning experimental procedures: setting the flow of the experiment, including practice test, key test and full attention test, recording whether the observer noticed the unexpected object in each test, and requiring to describe the characteristics thereof; S5, data collection and analysis: collecting key data, using statistical analysis method to analyze the difference in awareness rate under different experimental conditions, verifying the influence of saliency characteristics and attention set on involuntary blindness, and analyzing the interaction; S6, verification and optimization: comparing the experimental results with the theoretical assumptions, verifying whether it meets the expectations, adjusting the experimental design according to the results, optimizing the analysis method, and repeating the experiment to verify the stability of the results; S7, summarizing the main influencing factors of involuntary blindness, and putting forward an intelligent analysis framework based on saliency characteristics of involuntary blindness for predicting the awareness possibility of unexpected objects. 2.The intelligent analysis method based on saliency features of unintentional blindness according to claim 1, wherein, In S5, key data is collected, including: Awareness rate: whether the observer noticed the unexpected object; Error rate of counting task: evaluating the performance of the observer in the attention task. 3.The intelligent analysis method based on saliency features of unintentional blindness according to claim 2, wherein, In S7, the main influencing factors of involuntary blindness are summarized, including: bottom-up stimulus saliency, top-down attention set, and semantic characteristics of stimuli. 4.The method of claim 3, wherein, In S4, the practice test makes the observer familiar with the task, the key test introduces the unexpected object, and whether it is noticed is observed, and the full attention test does not set the attention task, and whether the unexpected object can be noticed is observed. 5.The method of claim 4, wherein, In S3, the experimental design only considers the similarity between the unexpected object and the attention item, and does not consider the similarity or dissimilarity between the unexpected object and the ignored item. 6.The method of claim 5, wherein, In S3, the experimental procedure follows the dynamic involuntary blindness paradigm, combines the dynamic nature of the selective fixation paradigm and the strict control of the static involuntary blindness paradigm, systematically manipulates the feature relationship between the attention item and the unexpected object in the attention set related dimensions and irrelevant dimensions through a series of experiments, explores the relationship between attention and visual perception processing, and the relationship between top-down and bottom-up perception processing, and investigates the specific conditions and influencing factors of involuntary blindness. 7.The method of claim 6, wherein, The experimental conditions also include observers and stimulus materials, the observers include: 43 observers with normal or corrected normal vision participate in the experiment, the observers are divided into groups of 6-8 people for testing, the data of 2 observers are subsequently excluded because the counting data is incomplete, no new object is found in the full attention test, according to the feature relationship between the unexpected object and the attention item, 41 observers are randomly allocated to 4 experimental conditions. 8.The method of claim 7, wherein, The stimuli consisted of a collision demonstration program written in Microsoft Visual C++ 6.0 presented on a 14-inch monitor with a screen resolution of 1024 768 pixels, and a grey background with a size of 582 472 pixels 18 cm 15.12 cm, and the size of the objects within the frame was 25 25 pixels 1 cm 1 cm. The observer sat approximately 40 cm from the computer screen with no head restraint and was presented with a dynamic involuntary blindness task using the computer. There were 8 figures of different colours and shapes within the interface of the collision demonstration program that moved randomly at variable speeds along independent trajectories, hitting the edges of the frame at irregular intervals and then bouncing smoothly back. The objects also collided with each other, changed direction and hit the edges of the frame. Each trial lasted 15 seconds. In each trial, the observer was asked to count the number of times a specific shape hit the edge of the frame. In the critical trials, an unexpected object, either a cyan or red-orange diamond, appeared. Each set of objects hit the edge of the frame 30 times in the demonstration interface to ensure strict control over the number of edge hits under different experimental conditions. The RGB values for each colour were as follows: blue RGB: 0, 0, 255, green RGB: 0, 255, 0, cyan RGB: 112, 219, 147, and red-orange RGB: 255, 36, 0. 9.The method of claim 8, wherein, The demographic information is collected according to the experimental design, and the number of edge collisions, whether the unexpected object appears, and the detailed description of the new object are recorded. Whether the observer is aware of the appearance of the unexpected object is analyzed by the following method: when the observer thinks he has seen the new object or different object appear, but the description is insufficient, or the description is irrelevant information or irrelevant change, the object movement or color change, it is considered that they have not perceived the unexpected object, and they are unperceived; when the observer thinks he has seen the new object appear, and at least one correct information is mentioned in the description of the unexpected object, it is considered that the observer has perceived the unexpected object, and he is a perceived person. The experimental process adopts the previous dynamic involuntary blindness paradigm. The collision demonstration program is presented in each test. The experimental procedure is as follows: 8 objects move in the frame, each object has shape and color characteristics, and is divided into two groups according to shape characteristics. Each observer completes 6 tests, and the instructions of different tests are slightly different. The observer is required to count the number of edge collisions of the specified color or shape object silently, and does not need to watch other objects in the frame. The first two tests do not include unexpected objects, and the last three tests include unexpected stimuli, but the instructions change. In the key test, after the objects in the frame move for 5 seconds, an unexpected stimulus will straightly cross the center of the screen from the left side, and disappear on the right side. The whole process lasts for 5 seconds. After each test, the observer records the number of edge collisions on the questionnaire, answers whether he sees the new object appear, and describes the new object in detail. In the attention test, although the observer is not explicitly told that there will be a new object, the instructions at the end of the key test change the observer's expectation of the appearance of the unexpected object. The observer records the number of edge collisions after the test, and is prompted whether he sees the new object appear, and then describes it in detail on the questionnaire. In the full attention test, the instructions change, and the observer is not required to count the number of edge collisions of any object silently. He only needs to focus on watching the movement of the objects in the display frame. The observer can completely focus his attention on observing all the objects in the display frame, so he will find the unexpected object. Then the observer is asked whether he sees the new object appear. The full attention test controls the observer's understanding and operation according to the instructions, ensures that any failure to report the key stimulus in the key test will not be explained as unable to see the stimulus. If the observer does not find the new object in this test, it is considered that his attention is not concentrated, or he does not understand and operate according to the instructions. His data will be deleted. Therefore, the detection rate of the observer in the full attention test is 100%, and there is no record of the number of edge collisions. The data will not be included in the experimental data analysis.

10. An intelligent analysis system based on salient features of involuntary blindness, characterized in that, It comprises: a definition module for defining attention tasks and attention set; a control stimulus characteristic module for determining the characteristics of the unexpected object and controlling the saliency and similarity of the unexpected object; The experimental condition setting module is used for designing experimental conditions, controlling characteristics of relevant dimensions and irrelevant dimensions of attentional set, ensuring random allocation of experimental conditions, and avoiding interference of experimental order on results. The experimental flow planning module is used for setting experimental flow, including practice test, key test and full attention test, recording whether the observer notices the unexpected object in each test, and requiring to describe the characteristics thereof. The data acquisition and analysis module is used for acquiring key data, using statistical analysis method to analyze differences in awareness rate under different experimental conditions, verifying influence of significant characteristics and attentional set on involuntary blindness, and analyzing interaction. The verification and optimization module is used for comparing experimental results with theoretical assumptions, verifying whether it meets expectations, adjusting experimental design according to results, optimizing analysis method, and repeating experiments to verify stability of results.

Images