Facial recognition monitoring alarm device

Abstract

The invention discloses a facial recognition monitoring alarm device, which comprises a facial image acquisition module, a personnel recognition judgment module, a normal recognition analysis module, a secondary recognition processing module and a monitoring information display module, relates to the technical field of facial recognition alarm, and solves the problem that complete facial features are difficult to accurately extract. In order to solve the technical problems that a face is divided into a core area and an auxiliary area by adopting a multi-area division and dynamic weight adjustment strategy on the basis of face key points and a shielding detection result, and weights are dynamically distributed, the technical problem that the recognition rate is greatly reduced and even cannot be recognized due to lack of fine analysis and dynamic weight adjustment on feature areas is solved. Meanwhile, L2 normalization and a dynamic cosine similarity threshold are combined, fine matching and weighted summation are carried out on global features to calculate an overall score, misjudgment caused by local feature deviation is effectively avoided, and the matching accuracy is improved in a complex illumination and posture change scene.

Description

Technical Field

[0001] The present invention relates to the technical field of facial recognition alarm technology, in particular to a facial recognition monitoring alarm device. Background Art

[0002] With the continuous growth of security needs, facial recognition technology has been widely used in the surveillance field.

[0003] According to a patent application with publication number CN112364696B, a method and system for improving home security by using home surveillance video are disclosed, wherein the method includes: in response to a facial image being a pre-stored facial image, identifying the facial image and determining whether the facial image contains panic expression information; when the facial image contains panic information, extracting sound information and / or person movement information in the video frame sequence, and judging whether the person has abnormal behavior based on the facial image, sound information and / or person movement information; when the person has abnormal behavior, determining the alarm terminal corresponding to the judgment scene based on the correspondence between the pre-set judgment scene and the alarm terminal; and sending an alarm message to the determined alarm terminal.

[0004] Existing facial recognition surveillance systems primarily use cameras to capture facial images and match them against a pre-existing database to identify individuals. However, practical applications present numerous challenges. Traditional systems struggle to accurately extract complete facial features, lack refined analysis of feature regions, and lack dynamic weighting adjustments, resulting in significantly reduced recognition rates or even failure to identify individuals. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a facial recognition monitoring and alarm device, which solves the problems of difficulty in accurately extracting complete facial features, lack of refined analysis of feature areas and dynamic weight adjustment, resulting in a significant drop in recognition rate or even inability to recognize.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A facial recognition monitoring and alarm device, comprising:

[0007] The person identification and judgment module is used to obtain the facial image of the person to be identified transmitted by the facial image acquisition module, judge its integrity, generate a matching analysis signal or an incomplete analysis signal, and transmit the incomplete analysis signal to the normal identification and analysis module;

[0008] Analyze the matching analysis signal, obtain unregistered persons based on whether there is a facial record in the database, and generate abnormal monitoring information or normal monitoring information based on real-time feedback information, and transmit it to the monitoring information display module at the same time;

[0009] The normal recognition and analysis module is used to analyze the incomplete signal, compare the incomplete area with the threshold, generate a re-acquisition signal or a recognition and analysis signal, process the recognition and analysis signal, extract facial features of the unobstructed area, and match them with the database to obtain preliminary matching results. At the same time, it extracts global features and matches them with the preliminary matching results, screens the pre-selected results, and transmits them to the secondary recognition processing module;

[0010] The secondary recognition processing module is used to analyze the pre-selected matching results, establish a rectangular coordinate system with the unobstructed area, obtain the facial features and the coordinate system position of the pre-selected results, calculate the corresponding distance value and compare it with the preset distance value, and screen out the secondary analysis features;

[0011] Calculate the proportion of the number of secondary analysis features and compare it with the proportion threshold to generate abnormal monitoring information or secondary analysis signals. At the same time, combine the iris information of the person to be identified to identify the secondary analysis signal to generate abnormal or normal monitoring information, and transmit it to the monitoring information display module.

[0012] As a further solution of the present invention, it also includes a facial image acquisition module and a monitoring information display module;

[0013] The facial image acquisition module is used to collect the facial image of the identified person through a high-definition camera and transmit it to the person recognition and judgment module;

[0014] The monitoring information display module is used to display the acquired abnormal or normal monitoring information to the corresponding management personnel.

[0015] As a further solution of the present invention, the specific manner in which the personnel identification and judgment module generates a matching analysis signal or an incomplete analysis signal is as follows:

[0016] Obtain a facial image and determine its integrity. If the image is unobstructed and facial features can be fully identified, it is considered complete and further matched with the database to generate a matching analysis signal.

[0017] If there is occlusion and some features cannot be effectively recognized, it is considered incomplete, and an incomplete analysis signal is generated and transmitted to the normal recognition and analysis module.

[0018] As a further solution of the present invention, the specific manner in which the personnel identification and judgment module analyzes the matching analysis signal is as follows:

[0019] Determine whether there is a corresponding facial record in the database. If so, generate a normal monitoring signal. Otherwise, feed it back to the corresponding management personnel and receive real-time feedback information from the management personnel for comprehensive judgment.

[0020] If the real-time feedback information is a recorded person, it will be marked as a registered person and normal monitoring information will be generated. Conversely, if the real-time feedback information is a non-recorded person, it will be marked as a non-registered person and abnormal monitoring information will be generated. Both will be transmitted to the monitoring information display module at the same time.

[0021] As a further solution of the present invention, the specific manner in which the normal recognition and analysis module analyzes the incomplete signal is as follows:

[0022] The incomplete area of ​​the facial image is extracted and compared with the threshold. If the incomplete area is larger than the threshold, a re-acquisition signal is generated and transmitted back to the facial image acquisition module for secondary facial image acquisition. If the incomplete area is smaller than the threshold, it indicates that the facial image can be recognized, and a recognition analysis signal is generated and analyzed.

[0023] As a further solution of the present invention, the specific manner in which the normal recognition and analysis module analyzes the recognition and analysis signal is as follows:

[0024] The unobstructed area corresponding to the facial image is obtained, and the corresponding facial feature label is extracted and recorded as i, and i = 1, 2, ..., j, where j represents the type of facial feature. The obtained facial features are then matched with the feature database, and preliminary matching results are obtained by screening. At the same time, all facial features are fused to obtain global features and the global features are matched with the preliminary matching results as a whole to obtain pre-selected matching results, which are then transmitted to the secondary recognition processing module.

[0025] As a further solution of the present invention, the normal recognition and analysis module obtains the preselected matching result in the following specific manner:

[0026] Obtain global features and match them with facial features corresponding to the preliminary matching results. Label the facial features corresponding to the preliminary matching results as to-be-matched features n, where n = 1, 2, ..., m, where m represents the type of to-be-matched features. Then, vectorize the global features and to-be-matched features, and calculate the overall matching score between the global feature vector and the to-be-matched feature vector.

[0027] The obtained overall matching score is compared with the preset value, and the preliminary screening results with an overall matching score greater than the preset value are screened and recorded as pre-selected matching results.

[0028] As a further solution of the present invention, the specific manner in which the secondary recognition processing module analyzes the pre-selected matching results is as follows:

[0029] Obtain the preselected matching result and label it as a, where a=1, 2, ..., b, where b represents the type of the preselected matching result. Then obtain the unobstructed area corresponding to the facial image, and use the unobstructed area to generate a square area with the same side length. Then, use the center point of the square area as the origin, and establish a corresponding rectangular coordinate system with the horizontal and vertical axes of the origin as the x-axis and y-axis respectively. At the same time, obtain the position of facial feature i on the rectangular coordinate system and record it as A. i (x i ,y i ), similarly obtain the facial feature position corresponding to the preselected matching result b b (x b ,y b );

[0030] And match the corresponding positions of the two, according to the distance formula The distance between the two positions on the coordinate system is calculated and compared with the preset distance value. If the distance is less than the preset distance, it indicates that the two positions correspond. Otherwise, if the distance is greater than the preset distance, it indicates that the two positions do not correspond. The corresponding facial features are recorded as secondary analysis features. Similarly, all secondary analysis features are obtained.

[0031] As a further solution of the present invention, the specific manner in which the secondary identification processing module generates abnormal or normal monitoring information is as follows:

[0032] Calculate the ratio of the secondary analysis features to all facial features, and compare the obtained ratio with the ratio threshold. If the ratio is greater than the ratio threshold, it indicates that the person to be identified is abnormal, and abnormal monitoring information is generated. Conversely, if the ratio is less than the ratio threshold, a secondary analysis signal is generated.

[0033] Then the secondary analysis signal is analyzed to obtain the iris information corresponding to the person to be identified, and it is matched with the iris information corresponding to the pre-selected matching result. If there is a match, normal monitoring information is generated. If there is no match, abnormal monitoring information is generated, and both are transmitted to the monitoring information display module at the same time.

[0034] The present invention provides a facial recognition monitoring and alarm device. Compared with the prior art, it has the following advantages:

[0035] This method uses a multi-region partitioning and dynamic weight adjustment strategy to divide the face into core and auxiliary regions based on facial key points and occlusion detection results, and dynamically assigns weights. Furthermore, combined with L2 normalization and a dynamic cosine similarity threshold, it performs a refined match of global features and calculates an overall score using a weighted summation. This effectively avoids misjudgments caused by local feature deviations and improves matching accuracy in scenes with complex lighting and posture variations.

[0036] The secondary recognition processing module of the present invention verifies the position of features by constructing a rectangular coordinate system, calculates the proportion of feature quantity, and performs secondary matching based on iris information. This multi-level verification mechanism can reduce the error rate compared to single facial recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 This is a block diagram of the system principle of the present invention. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0039] Example 1

[0040] See also Figure 1 The present application provides a facial recognition monitoring and alarm device, which includes a facial image acquisition module, a personnel recognition and judgment module, a normal recognition and analysis module, a secondary recognition processing module and a monitoring information display module.

[0041] The facial image acquisition module is used to collect the facial image of the person to be identified and transmit it to the person identification and judgment module, where the facial image of the person is acquired through the installed high-definition camera.

[0042] Personnel identification and judgment module, which is used to identify and analyze the acquired facial image for the person to be identified, and to judge the facial image integrity. If the facial image is complete, and the facial image integrity means that there is no occlusion on the face and the facial features can be fully identified, it will be further matched with the database to generate a matching analysis signal. On the contrary, if the facial image is incomplete, which means that there is occlusion in the facial image, resulting in partial inability to be effectively identified, an incomplete analysis signal will be generated and transmitted to the normal recognition and analysis module at the same time;

[0043] The generated matching analysis signal is processed to determine whether there is a corresponding facial record in the database. If so, the corresponding person to be identified is marked as a registered person and a normal monitoring signal is generated. Otherwise, the corresponding person to be identified is marked as an unregistered person and fed back to the corresponding management personnel. The management personnel's real-time feedback information is received for comprehensive judgment;

[0044] If the real-time feedback information is a recorded person, and the recorded person here is indicated as registered but not entered into the system, it will be marked as a registered person and normal monitoring information will be generated. On the contrary, if the real-time feedback information is a non-recorded person, it will be marked as a non-registered person and abnormal monitoring information will be generated. Both will be transmitted to the monitoring information display module at the same time.

[0045] Normal recognition and analysis module, which is used to analyze the acquired incomplete signal, extract the incomplete area of ​​the facial image, and compare the extracted incomplete area with a threshold value. The specific value of the threshold value is set by the operator. If the incomplete area is larger than the threshold value, it means that the facial image cannot be recognized, and a re-acquisition signal is generated. At the same time, it is reversely transmitted to the facial image acquisition module for secondary facial image acquisition. For example, during the personnel identification process at the subway security checkpoint, passenger A wears a wide-brimmed hat and a mask to pass the security check. The normal recognition and analysis module detects that the occlusion area reaches 45% (>threshold 30%) and generates a re-acquisition signal. The system prompts "Please take off the hat and mask". After the passenger adjusts, the second acquisition is successful. After repair and matching, it is confirmed as a registered passenger and is released. If the incomplete area is smaller than the threshold value, it means that the facial image can be recognized and an identification analysis signal is generated;

[0046] Then, the generated recognition analysis signal is processed to obtain the unobstructed area corresponding to the facial image, and the corresponding facial feature label is extracted and recorded as i, and i=1, 2, ..., j, where j represents the type of facial feature, and the extraction of facial features here is extracted by a lightweight convolutional neural network algorithm. This is the prior art and will not be described in detail here. Then, the obtained facial features are matched with the feature database, and the feature database is composed of the facial features of the registered persons. The preliminary matching results are obtained by screening, and the matching process is to match all facial features with the feature database separately and obtain the corresponding matching results. At the same time, all facial features are fused to obtain global features, and the global features here are represented as overall facial features, and are matched with the preliminary matching results as a whole to obtain pre-selected matching results, which are then transmitted to the secondary recognition processing module, and the specific matching method is as follows:

[0047] Obtain global features, and match the global features with facial features corresponding to the preliminary matching results. The facial features here represent features in the same area as the fused features. For example, if the global feature is a feature of one-half of the face (the upper half of the face), then the corresponding one-half feature (the upper half of the face) in the preliminary matching result is obtained. The facial features corresponding to the preliminary matching results are labeled as features to be matched n, and n=1, 2, ..., m, where m represents the type of features to be matched. Then, the global features and the features to be matched are vectorized, and the vectorization process is specifically represented by representing them through vectors. At the same time, the overall matching score between the global feature vector and the feature vector to be matched is calculated.

[0048] Based on facial key point detection technology (such as MTCNN positioning 68 key points), the face is divided into multiple semantic sub-regions as follows:

[0049] Core areas: eyes (including eye area), nose, and mouth;

[0050] Auxiliary areas: cheeks, forehead, jaw;

[0051] The system dynamically adjusts the region priority based on the occlusion detection results (such as DeepLabv3+ identifying masks, sunglasses, and other obstructions). For example, when a mask is detected, the weight of the mouth area is automatically reduced, the weight of the eyes and nose is increased to 0.4, and the weight of the cheeks is set to 0.2;

[0052] A lightweight convolutional neural network (such as MobileFaceNet) is used to extract the overall features of each region to obtain global features (facial features) and output a 128-dimensional feature vector to capture macro information such as facial contour and skin color. The global feature vector is L2 normalized to ensure that the vector modulus is 1 and eliminate scale differences. At the same time, the dynamic cosine similarity threshold is set according to the stability of the facial region. For example, the eye region is characterized by stability, and the threshold is set to 0.75. The nose region is less affected by expression, and the threshold is set to 0.70. The cheek allows a certain amount of feature difference, and the threshold is set to 0.65.

[0053] Then calculate the cosine similarity value of each facial feature and perform weighted summation according to the facial feature weight to obtain the overall matching score. The specific weighted summation formula is: Calculate the overall matching score S, where ω i is the weight of the i-th facial feature, cos i is the cosine similarity value of the i-th facial feature, j is the total number of facial features;

[0054] The obtained overall matching score is compared with a preset value, and the specific value of the preset value is set by the operator, and the preliminary screening results with an overall matching score greater than the preset value are screened and recorded as pre-selected matching results, and are transmitted to the secondary recognition processing module at the same time.

[0055] Example 2

[0056] As the second embodiment of the present invention, it is implemented on the basis of the first embodiment, and differs from the first embodiment in the following aspects:

[0057] The secondary recognition processing module is used to analyze the obtained pre-selected matching results and label them as a, where a=1, 2, ..., b, where b represents the type of the pre-selected matching result. Then, the unobstructed area corresponding to the facial image is obtained and a corresponding rectangular coordinate system is established based on the unobstructed area. The specific establishment method is as follows:

[0058] Taking the unobstructed area as the upper half of the face as an example for analysis, a square area with the same side length is generated from the unobstructed area, and the specific value of the side length of the square area here is set by the operator. The specific side length of the square is greater than the length of the upper half of the face. Then, the center point of the square area is taken as the origin, and the horizontal and vertical directions of the origin are used as the x-axis and y-axis to establish a corresponding rectangular coordinate system. At the same time, the position of the facial feature i on the rectangular coordinate system is obtained and recorded as A i (x i ,y i ), similarly obtain the facial feature position corresponding to the preselected matching result b b (x b ,y b ), and match the corresponding positions of the two, calculate the distance between the two in the coordinate system, and the distance value here is calculated by the distance formula between the two points, specifically At the same time, the calculated distance value is compared with a preset distance value, and the specific value of the preset distance value is set by the operator. If the distance value is less than the preset distance value, it means that there is a correspondence between the two positions. Conversely, if the distance value is greater than the preset distance value, it means that there is no correspondence between the two positions, and the corresponding facial features are recorded as secondary analysis features. And so on, all secondary analysis features are obtained;

[0059] Calculate the ratio of the number of secondary analysis features to all facial features, and compare the obtained ratio with the ratio threshold. The specific value of the ratio threshold is set by the operator. If the ratio is greater than the ratio threshold, it indicates that the person to be identified is abnormal and generates abnormal monitoring information. On the contrary, if the ratio is less than the ratio threshold, a secondary analysis signal is generated.

[0060] Then the secondary analysis signal is analyzed to obtain the iris information corresponding to the person to be identified, and it is matched with the iris information corresponding to the pre-selected matching result. If there is a match, normal monitoring information is generated. If there is no match, abnormal monitoring information is generated, and both are transmitted to the monitoring information display module at the same time.

[0061] The monitoring information display module is used to display the acquired normal monitoring information and abnormal monitoring information to the corresponding management personnel.

[0062] Example 3

[0063] As the third embodiment of the present invention, the focus is on combining the implementation processes of the first and second embodiments.

[0064] Some of the data in the above formulas are calculated based on their numerical values ​​and are not substituted into parameter units for calculation. At the same time, the contents not described in detail in this specification belong to the existing technology known to those skilled in the art.

[0065] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A facial recognition monitoring and alarm device, characterized in that: include: The person identification and judgment module is used to obtain the facial image of the person to be identified transmitted by the facial image acquisition module, judge its integrity, generate a matching analysis signal or an incomplete analysis signal, and transmit the incomplete analysis signal to the normal identification and analysis module; Analyze the matching analysis signal, obtain unregistered persons based on whether there is a facial record in the database, and generate abnormal monitoring information or normal monitoring information based on real-time feedback information, and transmit it to the monitoring information display module at the same time; The normal recognition and analysis module is used to analyze the incomplete signal, compare the incomplete area with the threshold, generate a re-acquisition signal or a recognition and analysis signal, process the recognition and analysis signal, extract facial features of the unobstructed area, and match them with the database to obtain preliminary matching results. At the same time, it extracts global features and matches them with the preliminary matching results, screens the pre-selected results, and transmits them to the secondary recognition processing module; The secondary recognition processing module is used to analyze the pre-selected matching results, establish a rectangular coordinate system with the unobstructed area, obtain the facial features and the coordinate system position of the pre-selected results, calculate the corresponding distance value and compare it with the preset distance value, and screen out the secondary analysis features; Calculate the proportion of the number of secondary analysis features and compare it with the proportion threshold to generate abnormal monitoring information or secondary analysis signals. At the same time, combine the iris information of the person to be identified to identify the secondary analysis signal to generate abnormal or normal monitoring information, and transmit it to the monitoring information display module.

2. A facial recognition monitoring and alarm device according to claim 1, characterized in that: It also includes a facial image acquisition module and a monitoring information display module; The facial image acquisition module is used to collect the facial image of the identified person through a high-definition camera and transmit it to the person recognition and judgment module; The monitoring information display module is used to display the acquired abnormal or normal monitoring information to the corresponding management personnel.

3. A facial recognition monitoring and alarm device according to claim 1, characterized in that: The specific method for the personnel identification and judgment module to generate a matching analysis signal or an incomplete analysis signal is: Obtain a facial image and determine its integrity. If the image is unobstructed and facial features can be fully identified, it is considered complete and further matched with the database to generate a matching analysis signal. If there is occlusion and some features cannot be effectively recognized, it is considered incomplete, and an incomplete analysis signal is generated and transmitted to the normal recognition and analysis module.

4. The facial recognition monitoring and alarm device according to claim 1, characterized in that: The specific method in which the personnel identification and judgment module analyzes the matching analysis signal is as follows: Determine whether there is a corresponding facial record in the database. If so, generate a normal monitoring signal. Otherwise, feed it back to the corresponding management personnel and receive real-time feedback information from the management personnel for comprehensive judgment. If the real-time feedback information is a recorded person, it will be marked as a registered person and normal monitoring information will be generated. Conversely, if the real-time feedback information is a non-recorded person, it will be marked as a non-registered person and abnormal monitoring information will be generated. Both will be transmitted to the monitoring information display module at the same time.

5. The facial recognition monitoring and alarm device according to claim 1, characterized in that: The specific method in which the normal recognition and analysis module analyzes the incomplete signal is as follows: The incomplete area of ​​the facial image is extracted and compared with the threshold. If the incomplete area is larger than the threshold, a re-acquisition signal is generated and transmitted back to the facial image acquisition module for secondary facial image acquisition. If the incomplete area is smaller than the threshold, it indicates that the facial image can be recognized, and a recognition analysis signal is generated and analyzed.

6. The facial recognition monitoring and alarm device according to claim 5, characterized in that: The specific way in which the normal recognition and analysis module analyzes the recognition and analysis signal is as follows: The unobstructed area corresponding to the facial image is obtained, and the corresponding facial feature label is extracted and recorded as i, and i = 1, 2, ..., j, where j represents the type of facial feature. The obtained facial features are then matched with the feature database, and preliminary matching results are obtained by screening. At the same time, all facial features are fused to obtain global features and the global features are matched with the preliminary matching results as a whole to obtain pre-selected matching results, which are then transmitted to the secondary recognition processing module.

7. The facial recognition monitoring and alarm device according to claim 6, characterized in that: The specific method for the normal recognition and analysis module to obtain the preselected matching result is: Obtain global features and match them with facial features corresponding to the preliminary matching results. Label the facial features corresponding to the preliminary matching results as to-be-matched features n, where n = 1, 2, ..., m, where m represents the type of to-be-matched features. Then, vectorize the global features and to-be-matched features, and calculate the overall matching score between the global feature vector and the to-be-matched feature vector. The obtained overall matching score is compared with the preset value, and the preliminary screening results with an overall matching score greater than the preset value are screened and recorded as pre-selected matching results.

8. The facial recognition monitoring and alarm device according to claim 1, characterized in that: The specific method in which the secondary recognition processing module analyzes the pre-selected matching results is as follows: Obtain the preselected matching result and label it as a, where a=1, 2, ..., b, where b represents the type of the preselected matching result. Then obtain the unobstructed area corresponding to the facial image, and use the unobstructed area to generate a square area with the same side length. Then, use the center point of the square area as the origin, and establish a corresponding rectangular coordinate system with the horizontal and vertical axes of the origin as the x-axis and y-axis respectively. At the same time, obtain the position of facial feature i on the rectangular coordinate system and record it as A. i (x i ,y i ), similarly obtain the facial feature position corresponding to the preselected matching result b b (x b ,y b ); And match the corresponding positions of the two, according to the distance formula The distance between the two positions on the coordinate system is calculated and compared with the preset distance value. If the distance is less than the preset distance, it indicates that the two positions correspond. Otherwise, if the distance is greater than the preset distance, it indicates that the two positions do not correspond. The corresponding facial features are recorded as secondary analysis features. Similarly, all secondary analysis features are obtained.

9. The facial recognition monitoring and alarm device according to claim 1, characterized in that: The specific method for the secondary identification processing module to generate abnormal or normal monitoring information is as follows: Calculate the ratio of the secondary analysis features to all facial features, and compare the obtained ratio with the ratio threshold. If the ratio is greater than the ratio threshold, it indicates that the person to be identified is abnormal, and abnormal monitoring information is generated. Conversely, if the ratio is less than the ratio threshold, a secondary analysis signal is generated. Then the secondary analysis signal is analyzed to obtain the iris information corresponding to the person to be identified, and it is matched with the iris information corresponding to the pre-selected matching result. If there is a match, normal monitoring information is generated. If there is no match, abnormal monitoring information is generated, and both are transmitted to the monitoring information display module at the same time.

Images