WO2023231479A1 - 瞳孔检测方法、装置、存储介质及电子设备 - Google Patents

瞳孔检测方法、装置、存储介质及电子设备 Download PDF

Info

Publication number
WO2023231479A1
WO2023231479A1 PCT/CN2023/078539 CN2023078539W WO2023231479A1 WO 2023231479 A1 WO2023231479 A1 WO 2023231479A1 CN 2023078539 W CN2023078539 W CN 2023078539W WO 2023231479 A1 WO2023231479 A1 WO 2023231479A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
images
detected object
face
human eye
Prior art date
Application number
PCT/CN2023/078539
Other languages
English (en)
French (fr)
Inventor
李元景
李荐民
王继生
陈涛
王加宝
Original Assignee
同方威视科技江苏有限公司
同方威视技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 同方威视科技江苏有限公司, 同方威视技术股份有限公司 filed Critical 同方威视科技江苏有限公司
Publication of WO2023231479A1 publication Critical patent/WO2023231479A1/zh

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/18—Eye characteristics, e.g. of the iris
    • G06V40/193—Preprocessing; Feature extraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/761—Proximity, similarity or dissimilarity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation

Definitions

  • the present disclosure relates to the field of image processing, and specifically to a pupil detection method, device, storage medium and electronic equipment.
  • Embodiments of the present disclosure provide a pupil detection method, device, storage medium, and electronic equipment to at least partially solve the technical problems existing in related technologies.
  • a pupil detection method including: detecting position information of a detected object, and obtaining multiple scene images of the detected object after determining that the detected object enters the target area according to the position information. ; Perform face recognition processing on multiple scene images to obtain multiple face images; detect human eyes in multiple face images to obtain multiple human eye images; detect pupils in multiple human eye images to obtain multiple human eye images. Information about the pupil of the human eye image.
  • the method further includes: after detecting pupils in multiple human eye images and obtaining pupil information of multiple human eye images, performing deduplication processing on multiple human face images.
  • a detected object retains a face image; determining the pupil information in the retained face image is the pupil detection result of the detected object.
  • the method further includes: after determining that the retained pupil information in the face image is the pupil detection result of the detected object, outputting the information of the pupil of the detected object and the detected object's pupil information. Face images.
  • the method further includes: before detecting the pupils in the multiple human eye images and obtaining the pupil information of the multiple human eye images, performing quality assessment on the multiple human eye images; detecting the multiple human eye images.
  • Pupils in human eye images are obtained to obtain pupil information of multiple human eye images, including: detecting pupils in human eye images that pass quality assessment in multiple human eye images, and obtaining pupils in human eye images that pass quality assessment. Information.
  • obtaining multiple scene images of the detected object includes: when a detected object appears in the target area, obtaining the location information of the detected object in the target area; if based on the obtained location information of the detected object The position information determines that the detected object is moving, reduces the exposure time of the image acquisition device used to obtain the scene image to the preset exposure time, and shoots the target area through the image acquisition device to obtain multiple scene images of the target area, where, The preset exposure time is not longer than the first preset duration.
  • the method further includes: performing an image capture on the target area through an image acquisition device. When shooting, fill in the target area.
  • the target area is an area that is no more than a depth-of-field range of the image capture device that captures multiple scene images.
  • detecting the position information of the detected object includes: detecting the position information of the detected object through a ranging sensor, wherein the ranging sensor and the image acquisition device are located on the same vertical plane.
  • the image acquisition device at least includes: at least two image acquisition devices distributed longitudinally on a vertical plane, and the imaging areas of the at least two image acquisition devices overlap.
  • performing deduplication processing on multiple face images and retaining one face image for one detected object includes: acquiring multiple face images in sequence, and acquiring the face image each time Finally, compare the image quality and face similarity between the current face image and the previous face image. If the face similarity between the current face image and the previous face image is higher than the similarity threshold, retain the current face image. The face image with high image quality in the previous face image will be output until the acquired face images reach the preset number and the retained face image will be output.
  • the current face image is similar to the previous face image
  • the degree is not higher than the similarity threshold or the difference between the shooting time of the current face image and the shooting time of the previous face image is not less than the second preset duration, and the current face image is output as the face image of the detected object.
  • a pupil detection device including: an acquisition module configured to detect position information of a detected object, and obtain the detected object after determining that the detected object enters the target area according to the position information. Multiple scene images of the object; the face recognition module is configured to perform face recognition processing on the multiple scene images to obtain multiple face images; the human eye detection module is configured to detect people in the multiple face images eyes to obtain multiple human eye images; the pupil detection module is configured to detect pupils in multiple human eye images and obtain pupil information of multiple human eye images.
  • an electronic device including: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the executable instructions via Execute the instructions to execute any pupil detection method provided by the embodiments of the present disclosure.
  • a computer-readable storage medium is also provided, on which a computer program is stored.
  • the computer program is executed by a processor, any one of the pupil detection methods provided by the embodiments of the present disclosure is implemented.
  • a computer program product which when run on a computer causes the computer to execute any of the above pupil detection methods.
  • the pupil detection method, device, storage medium and electronic device of the embodiment of the present disclosure obtains an image containing the detected object by acquiring multiple scene images of the target area, and can acquire multiple images containing the same detected object, and sequentially obtain the image containing the detected object.
  • the face, eyes and pupils of the detected object are identified in the image, and a human eye image including the pupils is obtained, thereby achieving the purpose of obtaining a clear human eye image of the detected person.
  • by acquiring multiple scene images of the target area it can ensure that multiple images of the same detected object are acquired, which can effectively avoid the situation where the human eyes of the detected object cannot be recognized from the image due to poor quality of the acquired images. occurrence, it can be seen that the embodiments of the present disclosure improve the accuracy of pupil detection and reduce the false detection rate.
  • Figure 1 is a detection system architecture in a security inspection scenario according to an exemplary embodiment of the present disclosure
  • Figure 2 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure
  • Figure 3 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure
  • Figure 4 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure.
  • Figure 5 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure.
  • Figure 6 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure.
  • Figure 7 is a schematic diagram of a pupil detection system according to an exemplary embodiment of the present disclosure.
  • Figure 8 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure.
  • Figure 9 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure.
  • Figure 10 is a schematic structural diagram of a pupil detection device according to an exemplary embodiment of the present disclosure.
  • FIG. 11 is a schematic structural diagram of an electronic device according to an exemplary embodiment of the present disclosure.
  • Example embodiments will now be described more fully with reference to the accompanying drawings.
  • Example embodiments may, however, be embodied in various forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concepts of the example embodiments. To those skilled in the art.
  • Figure 1 is a detection system architecture in a security inspection scenario according to an exemplary embodiment of the present disclosure.
  • the detection system architecture involves a computer device 110, an image acquisition device 120 and a detected object 130.
  • Embodiments of the present disclosure may be implemented by a computer device 110.
  • the computer device 110 can obtain the image obtained from the detected object detection device 120, recognize the image, such as face recognition, human eye recognition, or even pupil recognition, and output the recognition result, where the computer device 110 can be
  • the server may also belong to a terminal device, etc., and this disclosure does not specifically limit this.
  • the detected object detection device 120 may include an image acquisition device (not shown in FIG.
  • the image acquisition device may acquire images of a stationary or moving object to be detected, or the image acquisition device may also collect images according to whether the object to be detected is
  • the exposure time is adjusted in a stationary or moving state, where the image acquisition device can be a device with an image acquisition function such as a camera, a camera, or a high-definition infrared camera, which is not specifically limited in this disclosure.
  • the detected object detection device 120 may also include a ranging sensor to detect the position of the detected object 130 in real time.
  • the ranging sensor may interact with the image acquisition device so that the image acquisition device detects the position of the detected object 130 in real time. The shooting is performed within a certain distance from the image collection device to ensure that the image collection device can capture a clearer image of the detected object 130 .
  • Figure 2 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 2, the method includes:
  • step S202 the position information of the detected object is detected, and after it is determined that the detected object enters the target area according to the position information, multiple scene images of the detected object are obtained;
  • the target area may be a designated area at the entrance of a public scene that requires security detection. People who enter the designated area may be captured through an image collection device, such as a camera, installed at the entrance. (an example of a detected object) performs continuous shooting to obtain multiple fields containing the detected object. scene image. In order to capture clearer images of the detected object as much as possible, the target area can be continuously photographed at preset time intervals to obtain multiple scene images containing the same detected object, so that in the subsequent process, the multiple scene images can be obtained. Select images with clearer objects to be detected.
  • the continuous acquisition of multiple scene images of the target area may include images or video streams containing human faces.
  • step S204 perform face recognition processing on multiple scene images to obtain multiple face images
  • existing face recognition also called portrait recognition or facial recognition
  • face recognition technology can be used to perform face recognition processing on multiple scene images to obtain multiple face images, that is, in the scene images Accurately calibrate the position and size of human faces.
  • step S206 human eyes in multiple face images are detected to obtain multiple human eye images
  • currently existing human eye detection methods can be used, such as human eye detection methods based on template matching, human eye detection methods based on prior experience, and human eye detection methods based on statistical theory to detect each frame.
  • human eye detection methods based on template matching
  • human eye detection methods based on prior experience based on prior experience
  • human eye detection methods based on statistical theory to detect each frame.
  • the position and size of human eyes in facial images After detecting the human eyes in each face image, the human eye images can be extracted from the face images to obtain multiple human eye images.
  • step S208 pupils in multiple human eye images are detected to obtain pupil information of multiple human eye images.
  • existing image recognition technology can be used to perform pupil detection on each human eye image to obtain pupil information in each human eye image, where the pupil information can include the size of the pupil and the shape of the pupil. Color and other information.
  • detecting pupils in multiple human eye images and obtaining pupil information of multiple human eye images may include:
  • the distance between the detected object and the image acquisition device is obtained. Based on the distance and the pupil image in the face image of the detected object, the size of the pupil of the detected object is calculated, and the size of the pupil of the detected object is obtained. Size Information. For example, the size of the pupil of the detected subject can be calculated based on the distance and the size of the pixels occupied by the pupil image of the detected subject in the face image of the detected subject.
  • the pupil detection method in the embodiment of the present disclosure obtains an image containing the detected object by acquiring multiple scene images of the target area. Multiple images containing the same detected object can be acquired, and the detected object is sequentially identified from the image.
  • the human face, human eye and pupil are obtained to obtain the human eye image including the pupil, thus achieving the purpose of obtaining a clear human eye image of the person being detected.
  • by acquiring multiple scene images of the target area it can ensure that multiple images of the same detected object are acquired, which can effectively avoid the situation where the human eyes of the detected object cannot be recognized from the image due to poor quality of the acquired images. occur.
  • the pupil detection method may further include:
  • the multiple face images are deduplicated and one face image is retained for one detected object;
  • the retained pupil information in the face image is determined to be the pupil detection result of the detected object.
  • the pupil information of multiple human eye images obtained by performing pupil detection on multiple human eye images may or may not belong to the same person.
  • the face image with the best object image quality In the exemplary embodiment, only one face image can be output for each detected object. Therefore, it is first necessary to determine the number of pupil detection obtained from multiple human eye images. Whether the pupil information of the human eye image belongs to the same person, and then determine the face image with the best quality among the face images belonging to the same person as a face image to be output.
  • an identity identifier can be assigned to each detected object to identify the face image of the detected object, that is, , the face images belonging to the same detected object are all identified with the identity of the detected object. Based on this, the corresponding face image of each detected object can be output according to the identity of the detected object, which can ensure that each detected object has a face image output and avoid the occurrence of missed detection of the detected object.
  • each detected object After detecting the pupil of the detected object in the human eye image, and then deduplicating multiple face images of the same detected object, it can be obtained on the basis of ensuring that each detected object has a valid pupil detection result.
  • the best face image of each detected object can improve the accuracy of pupil detection and reduce the false detection rate.
  • Figure 3 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 3, based on the method shown in Figure 2, the method may further include:
  • step S302 after it is determined that the retained pupil information in the face image is the pupil detection result of the detected object, the pupil information of the detected object and the face image of the detected object are output.
  • each detected object is assigned an identity identifier.
  • the face image of the detected object can be output together. , to present the detected pupil detection results more intuitively.
  • Figure 4 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 4, based on the method shown in Figure 2, the method may further include:
  • step S402 before detecting the pupils in the multiple human eye images and obtaining the pupil information of the multiple human eye images, perform quality assessment on the multiple human eye images;
  • the inventor of the present disclosure analyzed the human eye images in the scene images captured by the image acquisition device and found that most of the blurred eye patterns are caused by the lens of the image acquisition device being out of focus. Therefore, the embodiment of the present disclosure To address this issue, targeted screening was conducted when selecting quality assessment algorithms.
  • the embodiment of the present disclosure adopts a quality assessment method without a reference image, and uses an algorithm based on the Laplacian (Laplacian operator) to detect whether the human eye image is clear.
  • the human eye image can be converted into a grayscale image first, and the single-channel grayscale image can be convolved with the Laplacian convolution kernel to obtain a response map, and then the standard deviation value of the response map can be calculated.
  • the size of the difference can be used as an indicator to measure the clarity of the human eye image.
  • the larger the value of the standard deviation value the clearer the human eye image. On the contrary, it means the blurr the human eye image.
  • Based on the comparison between the standard deviation value and the preset sharpness threshold it is determined whether the human eye image is clear.
  • the above variance value can be calculated using the built-in Laplacian function of OpenCV (OpenCV is a cross-platform computer vision and machine learning software library released based on the Apache2.0 license (open source)). For example, the quality of human eye images of different definitions can be assessed and processed, and then manual analysis can be performed.
  • the threshold of clarity can be determined, and the system can filter out images with clarity lower than the threshold.
  • Human eye images retaining human eye images with a resolution higher than this threshold, can achieve the capture of clear human eye images, and provide a good data basis for subsequent identification of pupils from human eye images.
  • Evaluation and filtering out blurry images can reduce the number of images that the system needs to process in the subsequent process, which can reduce the system load to a certain extent and improve the system's rapid response performance.
  • machine learning in order to enable the system to adapt to scenes with strong ambient light, or to effectively perform human eye detection on detected objects wearing glasses, machine learning can also be used for model training to evaluate image quality.
  • the algorithm undergoes certain optimizations.
  • Detecting the pupils in the multiple human eye images and obtaining the pupil information of the multiple human eye images may include:
  • step S2082 the pupils in the human eye images that pass the quality evaluation among the plurality of human eye images are detected, and information about the pupils in the human eye images that pass the quality evaluation is obtained.
  • the pupil detection method of the embodiment of the present disclosure can be applied to fixed-point capture of the human eye image of the detected object.
  • the detected object normally walks to a certain By gazing at the preset image collection point (a designated position in the target area) for a certain period of time, such as 3-5 seconds, the image collection device can capture a clear human eye image of the object being detected.
  • the pupil detection method of the embodiment of the present disclosure can also be applied to capture the human eye image of the detected object while walking (the detected object can walk normally through the target area and gaze at the image collection during walking). Just install it once).
  • Figure 5 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 5, in this method, multiple scene images of the detected object are obtained. include:
  • step S2022 when a detected object appears in the target area, the location information of the detected object in the target area is obtained;
  • step S2024 if it is determined that the detected object is moving according to the acquired position information of the detected object, the exposure time of the image acquisition device used to acquire the scene image is reduced to the preset exposure time, and the target area is imaged through the image acquisition device.
  • the exposure time of the image acquisition device used to acquire the scene image is reduced to the preset exposure time, and the target area is imaged through the image acquisition device.
  • the first preset duration can be set according to actual needs. For example, it can be set according to the intensity of ambient light in the target area. The greater the ambient light intensity, the shorter the first preset duration can be set accordingly. Alternatively, it can also be set according to the movement of the detected object. If the detected object moves faster, the first preset duration can be set to be shorter accordingly.
  • the exposure time of the image acquisition device when it is determined that the detected object moves within the target area, can be reduced based on the current ambient light intensity of the target area. For example, multiple preset exposure times can be set in advance. The values correspond to different ambient light intensities respectively. Based on this, when the ambient light intensity of the target area is different, the exposure time of the image acquisition device can be reduced to different preset exposure times. In another exemplary embodiment, the preset exposure time can also be set to a fixed value, such as 5 ms, and the corresponding gain of the image acquisition device is set to 3.0.
  • the detected object in the target area continues to move through the area. Therefore, it is necessary to capture images of the continuously moving detected object.
  • the pupil detection method can also be applied to pupil detection of moving detected objects, which improves the scene applicability of the system.
  • Figure 6 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 6, based on the method shown in Figure 5, step S2024 may also include: When the image acquisition device photographs the target area, it fills the target area with light.
  • the motion blur caused by the movement of the detected object is offset by reducing the exposure time of the image acquisition device.
  • reducing the exposure time of the image acquisition device will lead to a reduction in the amount of light entering the image acquisition device. Therefore,
  • embodiments of the present disclosure can supplement the light of the target area when acquiring a scene image of the target area.
  • an infrared fill light can be used to supplement the light of the target area, thereby ensuring image acquisition. Consistency of device imaging brightness.
  • the fill light can be turned on before the image acquisition device acquires the image of the target area.
  • the infrared fill light can be controlled to fill in the target area, and if no detected object is detected for a period of time After the object appears within the target area, the infrared fill light is controlled to stop working and stop filling the target area.
  • the fill light can be periodically controlled to provide auxiliary light to the target area according to the period in which the image collection device collects images.
  • Figure 7 is a schematic diagram of a pupil detection system according to an exemplary embodiment of the present disclosure.
  • the system includes: an infrared camera 712 (an example of the above image acquisition device), an I/O module 714 , switch 716, infrared fill light 718 (an example of the above fill light), a host 720 with key equipment such as CPU, motherboard, hard disk, memory, etc. (an example of the above computer equipment), and a laser ranging sensor 722 (An example of the above ranging sensor), in which the infrared fill light 718 is connected to the host 720 through the I/O module 714 and the switch 716, and the infrared camera 712 and the laser ranging sensor 722 are connected to the host 720 through USB.
  • the pupil detection method of the disclosed embodiment can be executed by the host 720 .
  • the host 720 sends a signal to the infrared fill light 718 through the I/O module 714 and the switch 716, and controls the infrared fill light 718 to turn on or off to provide fill light when the infrared camera 712 shoots.
  • the host 720 obtains the laser ranging sensor through USB. 722 sends the ranging signal, and based on the ranging information, it is determined that when the detected object walks to the depth of field range of the infrared camera 712, it passes The USB sends a signal to the infrared camera 712 to control the infrared camera 712 to photograph the detected object. It should be noted that if the system of the embodiment of the present disclosure is used for fixed-point detection of detected objects, the system may not include the laser ranging sensor 722.
  • FIG 8 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 8, in this method, detecting the position information of the detected object may include:
  • step S20222 the position information of the detected object is detected through the ranging sensor, where the ranging sensor and the image acquisition device are located on the same vertical plane.
  • disposing the ranging sensor and the image acquisition device on the same vertical plane allows the ranging sensor to detect the distance between the detected object and the image acquisition device. Based on this, when the ranging sensor detects When the detected object moves to the depth of field range of the image acquisition device, controlling the image acquisition device to capture the scene image of the target area ensures that the image acquisition device captures a clear image of the detected object by capturing the target area in the focused state. Since the depth of field range of the image acquisition device is limited, the detection time for the same detected object is shorter, and during the detection process, most of the unclear human eye images have been filtered, so it can be detected to a certain extent. Reduce system load and improve system performance. In addition, in order to improve the clarity of the captured image, the image acquisition device may be provided as a high-definition infrared camera, and in order to improve the ranging accuracy, the ranging sensor may be provided as a high-frequency laser ranging sensor.
  • the target area is an area that is no more than a depth-of-field range of the image capture device that captures multiple scene images.
  • the position information of the detected object in the target area is determined by the ranging sensor, and the image acquisition device is controlled to capture the image of the detected object when the detected object moves to the depth of field range of the image acquisition device. It ensures that a clear facial image of the detected object is captured during the movement of the detected object, without the need for the detected object to stay in front of the image acquisition device, which can achieve faster detection, and for the detected object, a true sense of the Sensitive over-inspection can effectively avoid the missed detection of the detected object.
  • the clear facial image of the detected object can also improve the accuracy of the pupil detection of the detected object and reduce the misjudgment rate. , improving the overall performance of the system.
  • the pupil detection method of the embodiment of the present disclosure can be applied to airports, major event venues, subways, and other entrances to public places with security detection requirements, in scenarios that require high detection accuracy and false detection rates.
  • the image acquisition device may at least include:
  • At least two image acquisition devices are longitudinally distributed on the vertical plane, and the imaging areas of the at least two image acquisition devices overlap.
  • two or even more image acquisition devices may be longitudinally distributed, so that the two or more image acquisition devices.
  • the vertical spacing between the center points of the lenses of each image acquisition device satisfies the overlap of imaging areas between the lenses to completely cover the head areas of detected persons of different heights, allowing the system to support the effective detection of persons in different height ranges. Therefore, in the security inspection scenario, the passing efficiency of the inspected personnel is greatly improved and the deployment rate of inspection human eyes is reduced.
  • Figure 9 is a flow chart of a pupil detection method according to an exemplary embodiment of the present disclosure. As shown in Figure 9, in this method, multiple face images are deduplicated, and one detected object is retained.
  • a face image can include:
  • Step S2102 Acquire multiple face images in sequence. After each face image is acquired, compare the image quality and face similarity between the current face image and the previous face image. If the current face image is different from the previous face image, The face similarity between the face images is higher than the similarity threshold, and the face image with high image quality between the current face image and the previous face image is retained until the acquired face images reach the preset number and then output The retained face image, if the similarity between the current face image and the previous face image is not higher than the similarity threshold or the shooting time of the current face image is different from the previous face image. The difference in the shooting time of a face image is not less than the second preset time length, and the current face image is output as the face image of the detected object.
  • the second preset duration is used to determine whether the current face image and the previous face image belong to the same person. Specifically, if the difference between the shooting time of the current face image and the shooting time of the previous face image is greater than During the second preset time period, the current face image and the previous face image are considered not to belong to the same person; if the difference between the shooting time of the current face image and the shooting time of the previous face image is not greater than the second preset time, For a preset time, the current face image and the previous face image are considered to belong to the same person.
  • the setting of the second preset duration can be set based on the actual duration required to detect different detected objects, and the embodiment of the present disclosure does not specifically limit this.
  • the purpose of deduplicating face images is to filter the multiple face images collected during the entire face collection process, and finally output a face image with the best effect.
  • the quality score of the face image can be calculated based on the image quality assessment algorithm, and the quality of the face image can be measured based on the score.
  • step S2102 illustrates the processing flow of step S2102 through an example.
  • the face image of the detected object is deduplicated using a deduplication processing algorithm, where the detected object is person A and person B.
  • Algorithm input multiple face images of the same person
  • Algorithm output a face image with the best quality.
  • the input images are 22 continuously collected images (that is, including 10 images of person A and 12 images of person B).
  • the images are named image 1, image 2..., image 10 (person A), and image 11. , image 12,..., image 22 (character B);
  • image 2 is used to perform a similarity comparison with the current image cache. If the comparison result shows that image 1 and image 2 are of the same person, and the quality score of the current image, that is, image 2 (calculated according to the quality assessment algorithm ) is greater than the current cached image, that is, image 1, then use image 2 to update the current image cache, otherwise, image 1 is still retained and image 2 is discarded.
  • image 1 is still retained and image 2 is discarded.
  • image 10 the image of person A with the best image quality is stored in the image cache.
  • the maximum time interval threshold corresponding to the same detected object is defined internally in the algorithm. , for example, if the timestamp between image 11 and image 10 (that is, the time interval between capturing image 11 and image 10) is greater than the threshold, even if it is determined through face similarity comparison that Figure 11 and Figure 10 belong to the same detected object, still Image 11 and images input after FIG. 11 are processed as face images belonging to different detected objects from those in FIG. 10 .
  • the algorithm will determine whether the face image of the last detected object has been output, and if it has not been output, it will be output.
  • Figure 10 is a schematic structural diagram of a pupil detection device according to an exemplary embodiment of the present disclosure. As shown in Figure 10, the device 100 includes:
  • the acquisition module 102 is configured to detect the position information of the detected object, and obtain multiple scene images of the detected object after determining that the detected object enters the target area according to the position information;
  • the face recognition module 104 is configured to perform face recognition processing on multiple scene images to obtain multiple face images;
  • the human eye detection module 106 is configured to detect human eyes in multiple human face images and obtain multiple human eye images;
  • the pupil detection module 108 is configured to detect pupils in multiple human eye images and obtain pupil information of multiple human eye images.
  • the pupil detection device in the embodiment of the present disclosure can be used alone to detect pupils, or can be combined with a security inspection device to detect the pupils of people passing inspection in security inspection scenarios.
  • the pupil detection device may further include:
  • the deduplication processing module is configured to detect the pupils in multiple human eye images and obtain the pupil information of the multiple human eye images, then perform deduplication processing on the multiple face images, and retain one image for each detected object. face images;
  • the determining module is configured to determine that the retained pupil information in the face image is a pupil detection result of the detected object.
  • the pupil detection device may further include:
  • the first output module is configured to output the information of the pupils of the detected subject and the face image of the detected subject after determining that the retained pupil information in the face image is the pupil detection result of the detected subject.
  • the human eye detection device may further include:
  • a quality assessment module configured to perform quality assessment on the multiple human eye images before detecting the pupils in the multiple human eye images and obtaining the pupil information of the multiple human eye images;
  • the pupil detection module can also be configured to:
  • the acquisition module may also be configured as:
  • the detection object If it is determined that the detected object is moving based on the acquired position information of the detected object, reduce the exposure time of the image acquisition device used to acquire the scene image to the preset exposure time, and shoot the target area through the image acquisition device to obtain multiple images.
  • the pupil detection device may further include:
  • the light filling module is configured to fill the target area with light when the target area is photographed by the image acquisition device.
  • the target area is an area that is no more than a depth-of-field range of the image acquisition device that captures multiple scene images.
  • the acquisition module may also be configured as:
  • the position information of the detected object is detected through a ranging sensor, wherein the ranging sensor and the image acquisition device are disposed on the same vertical plane.
  • the image acquisition device may at least include:
  • At least two image acquisition devices are longitudinally distributed on the vertical plane, and the imaging areas of the at least two image acquisition devices overlap.
  • the deduplication processing module is also configured to: acquire multiple face images in sequence, and after each face image is acquired, compare the image quality of the current face image with the previous face image. And face similarity. If the face similarity between the current face image and the previous face image is higher than the similarity threshold, the face with high image quality between the current face image and the previous face image is retained. images until the acquired face images reach the preset number and the retained face images are output. If the similarity between the current face image and the previous face image is not higher than the similarity threshold or the shooting time of the current face image is different from the previous face image. The difference in the shooting time of one face image is not less than the second preset time length, and the current face image is output as the face image of the detected object.
  • An embodiment of the present disclosure also provides an electronic device, including: a processor; and a memory for storing the executable instructions of the processor; wherein the processor is configured to execute any pupil detection method provided by the embodiment of the present disclosure by executing the executable instructions.
  • Embodiments of the present disclosure also provide a computer-readable storage medium on which a computer program is stored.
  • the computer program is executed by a processor, any one of the pupil detection methods provided by the embodiments of the present disclosure is implemented.
  • Embodiments of the present disclosure also provide a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the pupil detection methods in the above embodiments.
  • electronic device 1100 is embodied in the form of a general computing device.
  • the components of the electronic device 1100 may include, but are not limited to: the above-mentioned at least one processing unit 1110, the above-mentioned at least one storage unit 1120, and a bus 1130 connecting different system components (including the storage unit 1120 and the processing unit 1110).
  • the storage unit stores program code, and the program code can be executed by the processing unit 1110, so that the processing unit 1110 performs various exemplary methods according to the present disclosure described in the "Example Method" section of this specification.
  • the processing unit 1110 may perform step S202 as shown in FIG. 2 to detect the location information of the detected object, and after determining that the detected object enters the target area according to the location information, obtain the detected object.
  • Multiple scene images of the object Step S204, perform face recognition processing on the multiple scene images, and obtain multiple face images
  • Step S206 detect human eyes in the multiple face images, and obtain multiple human face images.
  • Eye image; step S208 detect the pupils in the multiple human eye images, and obtain the pupil information of the multiple human eye images.
  • the storage unit 1120 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 11201 and/or a cache storage unit 11202, and may further include a read-only storage unit (ROM) 11203.
  • RAM random access storage unit
  • ROM read-only storage unit
  • Storage unit 1120 may also include a program/utility 11204 having a set of (at least one) program modules 11205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, Each of these examples, or some combination, may include the implementation of a network environment.
  • program/utility 11204 having a set of (at least one) program modules 11205 including, but not limited to: an operating system, one or more application programs, other program modules, and program data, Each of these examples, or some combination, may include the implementation of a network environment.
  • Bus 1130 may be a local area representing one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or using any of a variety of bus structures. bus.
  • Electronic device 1100 may also communicate with one or more external devices 1200 (e.g., keyboard, pointing device, Bluetooth device, etc.), may also communicate with one or more devices that enable a user to interact with electronic device 1100, and/or with Any device (eg, router, modem, etc.) that enables the electronic device 1100 to communicate with one or more other computing devices. This communication may occur through an input/output (I/O) interface 1150.
  • the electronic device 1100 may also communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet) through the network adapter 1160. As shown, network adapter 1160 communicates with other modules of electronic device 1100 via bus 1130 .
  • network adapter 1160 communicates with other modules of electronic device 1100 via bus 1130 .
  • a computer-readable storage medium is also provided, on which a program product capable of implementing the method described above in this specification is stored.
  • various aspects of the present disclosure can also be implemented in the form of a program product, which includes program code.
  • the program product When the program product is run on a terminal device When running, the program code is used to cause the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Example Method" section of this specification.
  • a program product for implementing the above method according to an embodiment of the present disclosure is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer.
  • a readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.
  • the program product may take the form of any combination of one or more readable media.
  • the readable medium may be a readable signal medium or a readable storage medium.
  • the readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: electrical connection with one or more conductors, portable disk, hard disk, random access memory (RAM), read only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave carrying readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above.
  • a readable signal medium may also be any readable medium other than a readable storage medium that can send, propagate, or transport the program for use by or in connection with an instruction execution system, apparatus, or device.
  • Program code embodied on a readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical cable, RF, etc., or any suitable combination of the foregoing.
  • Program code for performing operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, etc., as well as conventional procedural Programming language—such as "C" or a similar programming language.
  • the program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server execute on.
  • the remote computing device may be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device, such as provided by an Internet service. (business comes via Internet connection).
  • LAN local area network
  • WAN wide area network
  • the example embodiments described here can be implemented by software, or can be implemented by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, U disk, mobile hard disk, etc.) or on the network , including several instructions to cause a computing device (which may be a personal computer, a server, a mobile terminal, a network device, etc.) to execute a method according to an embodiment of the present disclosure.
  • a computing device which may be a personal computer, a server, a mobile terminal, a network device, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • Multimedia (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Human Computer Interaction (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Computation (AREA)
  • Ophthalmology & Optometry (AREA)
  • Databases & Information Systems (AREA)
  • Artificial Intelligence (AREA)
  • Image Processing (AREA)

Abstract

一种瞳孔检测方法、装置、存储设备及电子设备，涉及图像处理技术领域，用以解决相关技术中瞳孔检测结果准确性较低的问题。涉及的瞳孔检测方法包括：检测被检测对象的位置信息，在根据所述位置信息确定所述被检测对象进入目标区域内后，获取所述被检测对象的多幅场景图像（S202）；对所述多幅场景图像进行人脸识别处理，得到多幅人脸图像（S204）；检测所述多幅人脸图像中的人眼，得到多幅人眼图像（S206）；检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息（S208）。本方案可提高瞳孔检测准确性以及降低误检率。

Description

瞳孔检测方法、装置、存储介质及电子设备
相关申请的交叉引用
本公开要求于2022年06月01日提交的申请号为202210623020.9、名称为“瞳孔检测方法、装置、存储介质及电子设备”的中国专利申请的优先权，该中国专利申请的全部内容通过引用全部并入本文。
技术领域
本公开涉及图像处理领域，具体而言，涉及一种瞳孔检测方法、装置、存储介质及电子设备。
背景技术
目前，在一些有安全检测需求的公共场所的入口处，通常需要对过检人员的面部甚至瞳孔的大小进行检测。但目前已有的检测技术中往往由于采集的图像质量较差、图像的亮度差异较大以及场景适应性差等原因导致检测结果准确性较低。
发明内容
本公开实施例提供一种瞳孔检测方法、装置、存储介质及电子设备，以用于至少部分地解决相关技术中存在的技术问题。
根据本公开的第一个方面，提供了一种瞳孔检测方法，包括：检测被检测对象的位置信息，在根据位置信息确定被检测对象进入目标区域内后，获取被检测对象的多幅场景图像；对多幅场景图像进行人脸识别处理，得到多幅人脸图像；检测多幅人脸图像中的人眼，得到多幅人眼图像；检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息。
在本公开的一些实施例中，所述方法还包括：在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之后，对多幅人脸图像进行去重处理，针对一个被检测对象保留一幅人脸图像；确定保留的人脸图像中的瞳孔的信息为被检测对象的瞳孔检测结果。
在本公开的一些实施例中，所述方法还包括：在确定保留的人脸图像中的瞳孔的信息为被检测对象的瞳孔检测结果之后，输出被检测对象的瞳孔的信息以及被检测对象的人脸图像。
在本公开的一些实施例中，所述方法还包括：在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之前，对多幅人眼图像进行质量评估；检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息，包括：检测多幅人眼图像中通过质量评估的人眼图像中的瞳孔，得到通过质量评估的人眼图像中的瞳孔的信息。
在本公开的一些实施例中，获取被检测对象的多幅场景图像，包括：在目标区域有被检测对象出现时，获取目标区域内被检测对象的位置信息；若根据获取的被检测对象的位置信息确定被检测对象在移动，降低用以获取场景图像的图像采集装置的曝光时间至预设曝光时间，并通过图像采集装置对目标区域进行拍摄，得到多幅目标区域的场景图，其中，预设曝光时间不长于第一预设时长。
在本公开的一些实施例中，所述方法还包括：在通过图像采集装置对目标区域进行 拍摄时，对目标区域进行补光。
在本公开的一些实施例中，目标区域为距离拍摄多幅场景图像的图像采集装置不超过图像采集装置的景深范围的区域。
在本公开的一些实施例中，检测被检测对象的位置信息，包括：通过测距传感器检测被检测对象的位置信息，其中，测距传感器与图像采集装置位于同一竖直面上。
在本公开的一些实施例中，图像采集装置至少包括：在竖直面上纵向分布的至少两个图像采集装置，至少两个图像采集装置的成像区域重叠。
在本公开的一些实施例中，对多幅人脸图像进行去重处理，针对一个被检测对象保留一幅人脸图像，包括：依次获取多幅人脸图像，在每次获取到人脸图像后，对比当前人脸图像与上一幅人脸图像的图像质量以及人脸相似度，若当前人脸图像与上一幅人脸图像之间的人脸相似度高于相似度阈值，保留当前人脸图像与上一幅人脸图像中图像质量高的人脸图像，直至获取的人脸图像达预设数量后输出保留的人脸图像，若当前人脸图像与上一幅人脸图像相似度不高于相似度阈值或当前人脸图像的拍摄时间与上一幅人脸图像的拍摄时间之差不小第二预设时长，将当前人脸图像作为被检测对象的人脸图像进行输出。
根据本公开的第二个方面，还提供了一种瞳孔检测装置，包括：获取模块，被配置为检测被检测对象的位置信息，在根据位置信息确定被检测对象进入目标区域内，获取被检测对象的多幅场景图像；人脸识别模块，被配置为对多幅场景图像进行人脸识别处理，得到多幅人脸图像；人眼检测模块，被配置为检测多幅人脸图像中的人眼，得到多幅人眼图像；瞳孔检测模块，被配置为检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息。
根据本公开的第三个方面，还提供了一种电子设备，包括：处理器；以及存储器，用于存储所述处理器的可执行指令；其中，所述处理器配置为经由执行所述可执行指令来执行本公开实施例提供的任意一种瞳孔检测方法。
根据本公开的第四个方面，还提供了一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现本公开实施例提供的任意一种瞳孔检测方法。
根据本公开的第五个方面，还提供了一种计算机程序产品，当其在计算机上运行时，使得计算机执行上述任意一种瞳孔检测方法。
本公开实施例的瞳孔检测方法、装置、存储介质及电子设备，通过获取目标区域的多幅场景图像来获取包含被检测对象的图像，可获取包含同一被检测对象的多幅图像，依次从该图像中识别出被检测对象的人脸、人眼以及瞳孔，得到包含瞳孔的人眼图像，从而实现了获取被检测人员清晰人眼图像的目的。其中，通过获取目标区域的多幅场景图像，可确保获取到同一被检测对象的多幅图像，可有效避免由于获取的图像质量不佳而导致无法从图像中识别到被检测对象人眼的情况发生，可见本公开实施例提高了瞳孔检测准确性，降低了误检率。
应当理解的是，以上的一般描述和后文的细节描述仅是示例性和解释性的，并不能限制本公开。
附图说明
此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本公开的实施例，并与说明书一起用于解释本公开的原理。显而易见地，下面描述中的附图仅仅是本公开的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是根据本公开一示例性实施例示出的一种在安检场景下的检测系统架构；
图2是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图3是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图4是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图5是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图6是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图7是根据本公开一示例性实施例示出的一种瞳孔检测系统的示意图；
图8是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图9是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图；
图10是根据本公开一示例性实施例示出的一种瞳孔检测装置的结构示意图；
图11是根据本公开一示例性实施例示出的一种电子设备的结构示意图。
具体实施方式
现在将参考附图更全面地描述示例实施方式。然而，示例实施方式能够以多种形式实施，且不应被理解为限于在此阐述的范例；相反，提供这些实施方式使得本公开将更加全面和完整，并将示例实施方式的构思全面地传达给本领域的技术人员。
此外，附图仅为本公开的示意性图解，并非一定是按比例绘制。图中相同的附图标记表示相同或类似的部分，因而将省略对它们的重复描述。附图中所示的一些方框图是功能实体，不一定必须与物理或逻辑上独立的实体相对应。可以采用软件形式来实现这些功能实体，或在一个或多个硬件模块或集成电路中实现这些功能实体，或在不同网络和/或处理器装置和/或微控制器装置中实现这些功能实体。
图1是根据本公开一示例性实施例示出的一种在安检场景下的检测系统架构，如图1所示，该检测系统架构涉及计算机设备110、图像采集装置120以及被检测对象130。本公开实施例可以由计算机设备110实现。其中，计算机设备110可以获取被检测对象检测装置120中获取图像，对该图像进行识别，例如进行人脸识别，人眼识别，甚至瞳孔识别，以及输出识别结果，其中，该计算机设备110可以是服务器，也可以属于终端设备等，本公开对此不做具体限制。被检测对象检测装置120中可包括图像采集装置(图1中暂未示出)，图像采集装置可以采集静止或运动中的被检测对象的图像，或者，图像采集装置还可根据被检测对象是处于静止或运动状态调整曝光时间，其中，该图像采集装置可以是相机、摄像头或高清红外相机等具有图像采集功能的装置，本公开对此不做具体限制。被检测对象检测装置120还可包括有测距传感器，用以对被检测对象130的位置进行实时检测，其中，测距传感器可以与图像采集装置进行数据交互，使得图像采集装置在被检测对象130距离图像采集装置一定距离范围内进行拍摄，以确保图像采集装置可以采集到被检测对象130较为清晰的图像。
需要说明的是，图1中所示的系统架构仅用于对本公开的方法进行示例性说明，并不对本公开具有任何限制性。本公开的方法可以应用于任何合适的系统架构中。
图2是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图2所示，该方法包括：
在步骤S202中，检测被检测对象的位置信息，在根据位置信息确定被检测对象进入目标区域内后，获取被检测对象的多幅场景图像；
在示例性实施例中，目标区域可以是有安全检测需求的公共场景所入口处的指定区域范围，可通过设置于该入口处的图像采集装置，如摄像头，对进入该指定区域范围内的人(为被检测对象的一个示例)进行连续拍摄，得到包含被检测对象的多幅场 景图像。为尽可能的捕捉到被检测对象较为清晰的图像，可以以预设时间间隔对目标区域进行连续拍摄，以获取包含同一被检测对象的多幅场景图像，以便于在后续过程中从该多幅图像中挑选出被检测对象较为清晰的图像。其中，连续获取目标区域的多幅场景图像可以包括含有人脸的图像或视频流。
在步骤S204中，对多幅场景图像进行人脸识别处理，得到多幅人脸图像；
在本公开的实施例中，可利用目前已有的人脸识别(也称人像识别或面部识别)技术对多幅场景图像进行人脸识别处理，得到多幅人脸图像，即在场景图像中准确标定出人脸的位置和大小。
在步骤S206中，检测多幅人脸图像中的人眼，得到多幅人眼图像；
在本公开的实施例中，可利用目前已有的人眼检测方式，如基于模板匹配的人眼检测方式、基于先验经验的人眼检测方式以及基于统计理论的人眼检测方式检测各幅人脸图像中人眼的位置和大小。在对各幅人脸图像中的人眼进行检测后，可从人脸图像中抠取人眼图像，得到多幅人眼图像。
在步骤S208中，检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息。
在本公开的实施例中，可利用目前已有的图像识别技术对各人眼图像进行瞳孔检测，以得到各人眼图像中瞳孔的信息，其中，瞳孔的信息可包括瞳孔的尺寸以及瞳孔的颜色等信息。
在示例性实施例中，检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息，可包括：
获取拍摄多幅人眼图像时，被检测对象距离图像采集装置的距离，根据该距离以及被检测对象人脸图像中的瞳孔图像，计算被检测对象的瞳孔的尺寸，得到被检测对象的瞳孔的尺寸信息。如，可基于该距离以及被检测对象的瞳孔图像在被检测对象的人脸图像中所占像素的小大，计算被检测对象的瞳孔的尺寸。
本公开实施例的瞳孔检测方法，通过获取目标区域的多幅场景图像来获取包含被检测对象的图像，可获取包含同一被检测对象的多幅图像，依次从该图像中识别出被检测对象的人脸、人眼以及瞳孔，得到包含瞳孔的人眼图像，从而实现了获取被检测人员清晰人眼图像的目的。其中，通过获取目标区域的多幅场景图像，可确保获取到同一被检测对象的多幅图像，可有效避免由于获取的图像质量不佳而导致无法从图像中识别到被检测对象人眼的情况发生。
在本公开的实施例中，瞳孔检测方法还可包括：
在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之后，对多幅人脸图像进行去重处理，针对一个被检测对象保留一幅人脸图像；
确定保留的人脸图像中的瞳孔的信息为被检测对象的瞳孔检测结果。
在本公开的实施例中，从多幅人眼图像进行瞳孔检测而得到的多幅人眼图像的瞳孔信息有可能属于同一人，也有可能不属于同一人，为了在检测过程中获得各被检测对象图像质量最佳的人脸图像，在示例性实施例中，针对每个被检测对象可仅输出一幅人脸图像，故，首先需确定从多幅人眼图像进行瞳孔检测而得到多幅人眼图像的瞳孔信息是否属于同一人，进而再确定出属于同一人的人脸图像中质量最佳的那幅人脸图像，作为待输出的一幅人脸图像。
在示例性实施例中，在对多幅场景图像进行人脸识别处理，得到人脸图像后，可为每一个被检测对象分配一个身份标识，用以标识该被检测对象的人脸图像，即，属于同一被检测对象的人脸图像均与该被检测对象的身份标识来标识。基于此，可根据被检测对象的身份标识为每个被检测对象输出其对应的人脸图像，可确保每个被检测对象均有人脸图像输出，避免有被检测对象漏检的情况发生。
在人眼图像中检测到被检测对象的瞳孔之后，再对同一被检测对象的多幅人脸图像进行去重处理，可在确保每个被检测对象均有有效瞳孔检测结果的基础上，得到每个被检测对象最佳的人脸图像，可提高瞳孔检测准确性、降低了误检率。
图3是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图3所示，该方法在图2所示的方法的基础上，还可进一步包括：
在步骤S302中，在确定保留的人脸图像中的瞳孔的信息为所述被检测对象的瞳孔检测结果之后，输出被检测对象的瞳孔的信息以及被检测对象的人脸图像。
如上文所述，在检测过程中，为每一个被检测对象分配了一个身份标识，在此基础上，在输出被检测对象瞳孔的尺寸信息时，可一并输出该被检测对象的人脸图像，以更加直观的呈现被检测的瞳孔检测结果。
图4是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图4所示，该方法在图2所示的方法的基础上，还可进一步包括：
在步骤S402中，在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之前，对多幅人眼图像进行质量评估；
本公开的发明人通过对图像采集装置拍摄得到的场景图像中的人眼图像进行分析发现，绝大部分的模糊眼图都是由于图像采集装置的镜头失焦导致的，因此，本公开实施例针对该问题在选择质量评估算法时进行了针对性的筛选。本公开实施例采用无参考图像的质量评估方式，使用基于Laplacian(拉普拉斯算子)的算法检测人眼图像是否清晰。可先将人眼图像转换为灰度图像，对单一通道的灰度图像与拉普拉斯卷积核进行卷积运算，从而得到一个响应图，再计算该响应图的标准差值，该标准差值的大小可以作为衡量人眼图像清晰度的指标，该标准差值的数值越大，表示人眼图像越清晰，反之，表示人眼图像越模糊。基于该标准差值与预先设定的清晰度的阈值进行比较，判断人眼图像是否清晰。其中，可采用OpenCV(OpenCV是一个基于Apache2.0许可(开源)发行的跨平台计算机视觉和机器学习软件库)内置的Laplacian函数进行计算得到上述方差值。示例性的，可通过对不同清晰度的人眼图像进行质量评估处理，再进行人工分析，结合系统对清晰度的要求，可确定清晰度的阈值，系统通过过滤清晰度掉低于该阈值的人眼图像，保留清晰度高于该阈值的人眼图像，可实现清晰人眼图像的捕获，为后续从人眼图像中识别瞳孔提供了良好的数据基础，同时，通过对人眼图像进行质量评估，过滤掉模糊图像，可减少系统在后续过程中需处理的图像的数量，可在一定程度上降低系统负载，提升系统的快速响应性能。
在示例性实施例中，为使系统能够适应环境光较强的场景下，或可有效对佩戴了眼镜的被检测对象进行人眼检测，也可采用机器学习进行模型训练来对图像的质量评估算法进行一定的优化。
检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息，可包括：
在步骤S2082中，检测多幅人眼图像中通过质量评估的人眼图像中的瞳孔，得到通过质量评估的人眼图像中的瞳孔的信息。
在示例性实施例中，本公开实施例的瞳孔检测方法可适用于对被检测对象的人眼图像进行定点捕捉，如，在对被检测对象进行瞳孔检测时，被检测对象正常行走至某一预先设定的图像采集点(目标区域中的指定位置)注视图像采集装置一定时长，如3-5s，图像采集装置即可捕捉到被检测对象清晰的人眼图像。
在示例性实施例中，本公开实施例的瞳孔检测方法还可适用于对行走中的被检测对象的人眼图像进行捕捉(被检测对象可正常行走通过目标区域，在行走过程中注视图像采集装置一下即可)，图5是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图5所示，在该方法中，获取被检测对象的多幅场景图像，可包括：
在步骤S2022中，在目标区域有被检测对象出现时，获取目标区域内被检测对象的位置信息；
在步骤S2024中，若根据获取的被检测对象的位置信息确定被检测对象在移动，降低用以获取场景图像的图像采集装置的曝光时间至预设曝光时间，并通过图像采集装置对目标区域进行拍摄，得到多幅目标区域的场景图，其中，预设曝光时间不长于第一预设时长。该第一预设时长可根据实际需求进行设置，如，可根据目标区域的环境光的强度进行设定，环境光强度越大，该第一预设时长可相应的被设置的越短。或者，还可根据被检测对象的移动情况进行设定，若被检测对象移动的越快，该第一预设时长可相应的被设置的越短。
在一个示例性实施例中，在确定被检测对象在目标区域内移动时，可结合目标区域当前的环境光强度，降低图像采集装置的曝光时间，如，可预先设置多个预设曝光时间的值，分别对应不同的环境光强度，基于此，可在目标区域的环境光强度不同时，将图像采集装置的曝光时间降低至不同的预设曝光时间。在另一个示例性实施例中，该预设曝光时间还可被设定为一固定值，如5ms，相应的图像采集装置的增益被设置为3.0。
在示例性实施例中，目标区域内的被检测对象持续移动通过该区域，故，需要对持续移动的被检测对象的图像进行捕捉，本公开实施例在判断出被检测对象在目标区域内移动时，通过将图像采集装置的曝光时间大幅度的调低来应对由于被检测对象移动而造成的运动模糊，从而可实现对移动中的被检测对象清晰人脸图像的获取，使得本公开实施例的瞳孔检测方法还可适用于对移动中的被检测对象的瞳孔检测，提高了系统的场景适用性。
图6是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图6所示，该方法在图5所示的方法的基础上，步骤S2024还可包括，在通过所述图像采集装置对所述目标区域进行拍摄时，对所述目标区域进行补光。
在图5所示的实施例中，通过降低图像采集装置的曝光时间来抵消被检测对象移动而造成的运动模糊，但图像采集装置的曝光时间降低会导致图像采集装置的进光量降低，故，为提高图像采集装置的进光量，本公开实施例在获取目标区域的场景图像时，可对目标区域进行补光，如，可采用红外补光灯对目标区域进行辅助补光，从而保证图像采集装置成像亮度的一致性。可在图像采集装置获取目标区域的图像之前即开启补光灯。此外，为降低能耗，可在检测到有被检测对象出现在目标区域的范围内时，控制红外补光灯工作，对目标区域进行补光，以及，在一段时间内未检测到有被检测对象出现在目标区域的范围内后，控制红外补光灯停止工作，停止对目标区域进行补光。又或者，可根据图像采集装置采集图像的周期，周期性的控制补光灯对目标区域进行辅助补光。
图7是根据本公开一示例性实施例示出的一种瞳孔检测系统的示意图，如图7所示，该系统包括：红外相机712(为上述图像采集装置的一个示例)、I/O模块714、交换机716、红外补光灯718(为上述补光灯的一个示例)、带有CPU、主板、硬盘、内存等关键设备的主机720(为上述计算机设备的一个示例)以及激光测距传感器722(为上述测距传感器的一个示例)，其中，红外补光灯718通过I/O模块714、经交换机716与主机720相连，红外相机712与激光测距传感器722通过USB与主机720相连，本公开实施例的瞳孔检测方法可由该主机720执行。主机720通过I/O模块714以及交换机716向红外补光灯718发送信号，控制红外补光灯718开启或关闭，以在红外相机712拍摄时进行补光，主机720通过USB获取激光测距传感器722发送的测距信号，根据该测距信息确定被检测对象行走至红外相机712景深范时，通过 USB向红外相机712发送信号，控制红外相机712对被检测对象进行拍摄。需要说明的是，若本公开实施例的系统应用于对被检测对象进行定点检测，该系统也可不包括激光测距传感器722。
图8是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图8所示，在该方法中，检测被检测对象的位置信息，可包括：
在步骤S20222中，通过测距传感器检测被检测对象的位置信息，其中，测距传感器与图像采集装置位于同一竖直面上。
在示例性实施例中，将测距传感器与图像采集装置设置于同一竖直面上，可使得测距传感器可检测出被检测对象距离图像采集装置的距离，进基于此，当测距传感器检测到被检测对象移动至图像采集装置的景深范围时，控制图像采集装置拍摄目标区域的场景图像，可确保图像采集装置在对焦状态下通过对目标区域的拍摄实现对被检测对象清晰图像的捕捉，由于图像采集装置景深范围有限，因此对于同一个被检测对象来说，其被检测时间较短，且在检测过程中，不清晰的人眼图像大部分已被过滤，故，在一定程度上可降低系统负载，提高系统性能。此外，为提高拍摄得到的图像的清晰度，图像采集装置可被提供为高清红外相机，为提高测距精度，测距传感器可被提供为高频率激光测距传感器。
在一些实施例中，目标区域为距离拍摄多幅场景图像的图像采集装置不超过图像采集装置的景深范围的区域。
在本公开的实施例中，通过测距传感器对被检测对象在目标区域内的位置信息，在被检测对象移动至图像采集装置的景深范围内，控制图像采集装置拍摄被检测对象的图像，可确保在被检测对象移动的过程中捕捉到被检测对象清晰的面部图像，无需被检测对象在图像采集装置前停留，可实现更加快速的检测，而对被检测对象来说，实现了真正意义上的无感知化过检，且，可有效避免有被检测对象漏检的情况发生，同时，被检测对象清晰的面部图像也可提高对被检测对象的瞳孔检测的准确性，降低了误判率，提高了系统综合性能。将该方法应用于安检场景下，可大幅提高被检人员的通行效率，大幅降低查验人员的配置率。基于此，本公开实施例的瞳孔检测方法可被应用于机场、重大赛事场馆、地铁及其他有安全检测需求的公共场所入口处等对检测准确率以及误检率要求较高的场景。
在本公开的实施例中，所述图像采集装置至少可包括：
在竖直面上纵向分布的至少两个图像采集装置，至少两个图像采集装置的成像区域重叠。
在示例性实施例中，为确保可检测到被检测对象的有效区域(即被检测对象的头部区域)，可采用两个甚至多个图像采集装置纵向分布的方式，使该两个或多个图像采集装置镜头的中心点垂直间距满足镜头之间的成像区域重叠，以完整覆盖不同身高的被检测人员的头部区域，使得系统支持不同身高区间范围内的人员的有效检测。因此，在安检场景下，大幅度提高了被检测人员的通行效率，降低了查验人眼的配置率。
图9是根据本公开一示例性实施例示出的一种瞳孔检测方法的流程图，如图9所示，在该方法中，对多幅人脸图像进行去重处理，针对一个被检测对象保留一幅人脸图像，可包括：
步骤S2102：依次获取多幅人脸图像，在每次获取到人脸图像后，对比当前人脸图像与上一幅人脸图像的图像质量以及人脸相似度，若当前人脸图像与上一幅人脸图像之间的人脸相似度高于相似度阈值，保留当前人脸图像与上一幅人脸图像中图像质量高的人脸图像，直至获取的人脸图像达预设数量后输出保留的人脸图像，若当前人脸图像与上一幅人脸图像相似度不高于相似度阈值或当前人脸图像的拍摄时间与上 一幅人脸图像的拍摄时间之差不小于第二预设时长，将当前人脸图像作为被检测对象的人脸图像进行输出。其中，第二预设时长用以判定当前人脸图像与上一幅人脸图像是否属于同一人，具体的，若当前人脸图像的拍摄时间与上一幅人脸图像的拍摄时间之差大于该第二预设时长，当前人脸图像与上一幅人脸图像被认为不属于同一人；若当前人脸图像的拍摄时间与上一幅人脸图像的拍摄时间之差不大于该第二预设时长，当前人脸图像与上一幅人脸图像被认为属于同一人。该第二预设时长的设定可基于检测不同被检测对象所需的实际时长来进行设定，本公开实施例对此不做具体限制。
其中，对人脸图像进行去重处理，是为了解决在整个人脸采集环节中，对于采集的多张人脸图像进行过滤，最终输出一张效果最佳的人脸图像。
在上述步骤S2102中，可基于图像质量评估算法计算得到人脸图像的质量得分，基于该得分衡量人脸图像的质量。
以下通过一个例子对步骤S2102的处理流程进行示例性说明，在该例子中，通过去重处理算法对被检测对象的人脸图像进行去重处理，其中，被检测对象以人物A以及人物B进行示例。
假设在人脸采集环境，对人物A采集到了10幅图像，对人物B采集到了12幅图像，那么去重处理的目标是，最终人物A，B分别会输出一幅人脸图像。
算法输入：同一个人的多幅人脸图像；
算法输出：一幅质量最佳的人脸图像。
假设输入图像为连续采集到的22幅图像(即包括人物A的10幅图像以及人物B的12幅图像)，图像依次被命名为图像1，图像2…，图像10(人物A)，图像11，图像12，…，图像22(人物B)；
第一幅图像到达时，使用图像1更新当前图像缓存(保留图像1)。
第二幅图像到达时，使用图像2与当前图像缓存进行相似度比对，如果对比结果表明图像1与图像2为同一人，并且当前图像，即图像2的质量分(根据质量评估算法计算出)大于当前缓存图像，即图像1，则使用图像2更新当前图像缓存，否则，仍保留图像1，舍弃图像2。依次按照该过程处理其他后续图像，直至图像10。此时图像缓存中存放的是人物A的图像质量最佳的图像。
在后续过程中，如果通过将图像11与图像10进行对比，发现图像11与图像10的相似度低于相似度阈值，确定图像11与图像10并非同一人的图像，此时，如果前10幅图像对应的人脸图像没有输出，则进行强制输出，即输出图像缓存中存放的人物A的最佳图像，并且更新当前缓存为图像11，根据上面的流程重复进行处理。
在去重处理算法中，为避免同一被检测对象经过连续两次的图像采集，仍没有该被检测对象的人脸图像输出的情况发生，算法内部定义了同一被检测对象对应的最大时间间隔阈值，例如，如果图像11与图像10之间的时间戳(即采集图像11与图像10的时间间隔)大于该阈值，即使通过人脸相似度对比确定图11与图10属于同一被检测对象，仍将图像11以及图11之后输入的图像作为与图10属于不同被检测对象的人脸图像进行处理。算法会判断上一个被检测对象的人脸图像是否已经输出，如果未输出，则输出。
图10是根据本公开一示例性实施例示出的一种瞳孔检测装置的结构示意图，如图10所示，该装置100包括：
获取模块102，被配置为检测被检测对象的位置信息，在根据位置信息确定被检测对象进入目标区域内，获取被检测对象的多幅场景图像；
人脸识别模块104，被配置为对多幅场景图像进行人脸识别处理，得到多幅人脸图像；
人眼检测模块106，被配置为检测多幅人脸图像中的人眼，得到多幅人眼图像；
瞳孔检测模块108，被配置为检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息。
可以理解的是，本公开实施例的瞳孔检测装置可单独用于对瞳孔进行检测，也可与安检装置结合，以在安检场景下实现对过检人员的对瞳孔进行检测。
在本公开的实施例中，瞳孔检测装置还可包括：
去重处理模块，被配置为在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之后，对多幅人脸图像进行去重处理，针对一个被检测对象保留一幅人脸图像；
确定模块，被配置为确定保留的人脸图像中的瞳孔的信息为被检测对象的瞳孔检测结果。
在本公开的实施例中，瞳孔检测装置还可包括：
第一输出模块，被配置为在确定保留的人脸图像中的瞳孔的信息为被检测对象的瞳孔检测结果之后，输出被检测对象的瞳孔的信息以及被检测对象的人脸图像。
在本公开的实施例中，人眼检测装置还可包括：
质量评估模块，被配置为在检测多幅人眼图像中的瞳孔，得到多幅人眼图像的瞳孔的信息之前，对多幅人眼图像进行质量评估；
瞳孔检测模块还可被配置为：
检测多幅人眼图像中通过质量评估的人眼图像中的瞳孔，得到通过质量评估的人眼图像中的瞳孔的信息。
在本公开的实施例中，获取模块还可被配置为：
在目标区域有被检测对象出现时，获取目标区域内被检测对象的位置信息；
若根据获取的被检测对象的位置信息确定被检测对象在移动，降低用以获取场景图像的图像采集装置的曝光时间至预设曝光时间，并通过图像采集装置对目标区域进行拍摄，得到多幅所述目标区域的场景图，其中，预设曝光时间不长于第一预设时长。
在本公开的实施例中，瞳孔检测装置还可包括：
补光模块，被配置为在通过图像采集装置对所述目标区域进行拍摄时，对所述目标区域进行补光。
在本公开的实施例中，目标区域为距离拍摄多幅场景图像的图像采集装置不超过图像采集装置的景深范围的区域。
在本公开的实施例中，所述获取模块还可被配置为：
通过测距传感器检测所述被检测对象的位置信息，其中，测距传感器与图像采集装置设置于同一竖直面上。
在本公开的实施例中，图像采集装置至少可包括：
在竖直面上纵向分布的至少两个图像采集装置，至少两个图像采集装置的成像区域重叠。
在本公开的实施例中，去重处理模块还被配置为：依次获取多幅人脸图像，在每次获取到人脸图像后，对比当前人脸图像与上一幅人脸图像的图像质量以及人脸相似度，若当前人脸图像与上一幅人脸图像之间的人脸相似度高于相似度阈值，保留当前人脸图像与上一幅人脸图像中图像质量高的人脸图像，直至获取的人脸图像达预设数量后输出保留的人脸图像，若当前人脸图像与上一幅人脸图像相似度不高于相似度阈值或当前人脸图像的拍摄时间与上一幅人脸图像的拍摄时间之差不小第二预设时长，将当前人脸图像作为所述被检测对象的人脸图像进行输出。
本公开实施例还提供了一种电子设备，包括：处理器；以及存储器，用于存储所 述处理器的可执行指令；其中，所述处理器配置为经由执行所述可执行指令来执行本公开实施例提供的任意一种瞳孔检测方法。
本公开实施例还提供了一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现本公开实施例提供的任意一种瞳孔检测方法。
本公开实施例还提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行上述实施例中任意一种瞳孔检测方法。
所属技术领域的技术人员能够理解，本公开的各个方面可以实现为系统、方法或程序产品。因此，本公开的各个方面可以具体实现为以下形式，即：完全的硬件实施方式、完全的软件实施方式(包括固件、微代码等)，或硬件和软件方面结合的实施方式，这里可以统称为“电路”、“模块”或“系统”。
如图11所示，电子设备1100以通用计算设备的形式表现。电子设备1100的组件可以包括但不限于：上述至少一个处理单元1110、上述至少一个存储单元1120、连接不同系统组件(包括存储单元1120和处理单元1110)的总线1130。
其中，所述存储单元存储有程序代码，所述程序代码可以被所述处理单元1110执行，使得所述处理单元1110执行本说明书上述“示例性方法”部分中描述的根据本公开各种示例性实施方式的步骤。例如，所述处理单元1110可以执行如图2中所示的步骤S202，检测被检测对象的位置信息，在根据所述位置信息确定所述被检测对象进入目标区域内后，获取所述被检测对象的多幅场景图像；步骤S204，对所述多幅场景图像进行人脸识别处理，得到多幅人脸图像；步骤S206，检测所述多幅人脸图像中的人眼，得到多幅人眼图像；步骤S208，检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息。
存储单元1120可以包括易失性存储单元形式的可读介质，例如随机存取存储单元(RAM)11201和/或高速缓存存储单元11202，还可以进一步包括只读存储单元(ROM)11203。
存储单元1120还可以包括具有一组(至少一个)程序模块11205的程序/实用工具11204，这样的程序模块11205包括但不限于：操作系统、一个或者多个应用程序、其它程序模块以及程序数据，这些示例中的每一个或某种组合中可能包括网络环境的实现。
总线1130可以为表示几类总线结构中的一种或多种，包括存储单元总线或者存储单元控制器、外围总线、图形加速端口、处理单元或者使用多种总线结构中的任意总线结构的局域总线。
电子设备1100也可以与一个或多个外部设备1200(例如键盘、指向设备、蓝牙设备等)通信，还可与一个或者多个使得用户能与该电子设备1100交互的设备通信，和/或与使得该电子设备1100能与一个或多个其它计算设备进行通信的任何设备(例如路由器、调制解调器等等)通信。这种通信可以通过输入/输出(I/O)接口1150进行。并且，电子设备1100还可以通过网络适配器1160与一个或者多个网络(例如局域网(LAN)，广域网(WAN)和/或公共网络，例如因特网)通信。如图所示，网络适配器1160通过总线1130与电子设备1100的其它模块通信。应当明白，尽管图中未示出，可以结合电子设备1100使用其它硬件和/或软件模块，包括但不限于：微代码、设备驱动器、冗余处理单元、外部磁盘驱动阵列、RAID系统、磁带驱动器以及数据备份存储系统等。
在本公开的示例性实施例中，还提供了一种计算机可读存储介质，其上存储有能够实现本说明书上述方法的程序产品。在一些可能的实施方式中，本公开的各个方面还可以实现为一种程序产品的形式，其包括程序代码，当所述程序产品在终端设备上 运行时，所述程序代码用于使所述终端设备执行本说明书上述“示例性方法”部分中描述的根据本公开各种示例性实施方式的步骤。
描述了根据本公开的实施方式的用于实现上述方法的程序产品，其可以采用便携式紧凑盘只读存储器(CD-ROM)并包括程序代码，并可以在终端设备，例如个人电脑上运行。然而，本公开的程序产品不限于此，在本文件中，可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。
所述程序产品可以采用一个或多个可读介质的任意组合。可读介质可以是可读信号介质或者可读存储介质。可读存储介质例如可以为但不限于电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。可读存储介质的更具体的例子(非穷举的列表)包括：具有一个或多个导线的电连接、便携式盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。
计算机可读信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了可读程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。可读信号介质还可以是可读存储介质以外的任何可读介质，该可读介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。
可读介质上包含的程序代码可以用任何适当的介质传输，包括但不限于无线、有线、光缆、RF等等，或者上述的任意合适的组合。
可以以一种或多种程序设计语言的任意组合来编写用于执行本公开操作的程序代码，所述程序设计语言包括面向对象的程序设计语言—诸如Java、C++等，还包括常规的过程式程序设计语言—诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算设备上执行、部分地在用户设备上执行、作为一个独立的软件包执行、部分在用户计算设备上部分在远程计算设备上执行、或者完全在远程计算设备或服务器上执行。在涉及远程计算设备的情形中，远程计算设备可以通过任意种类的网络，包括局域网(LAN)或广域网(WAN)，连接到用户计算设备，或者，可以连接到外部计算设备(例如利用因特网服务提供商来通过因特网连接)。
应当注意，尽管在上文详细描述中提及了用于动作执行的设备的若干模块或者单元，但是这种划分并非强制性的。实际上，根据本公开的实施方式，上文描述的两个或更多模块或者单元的特征和功能可以在一个模块或者单元中具体化。反之，上文描述的一个模块或者单元的特征和功能可以进一步划分为由多个模块或者单元来具体化。
此外，尽管在附图中以特定顺序描述了本公开中方法的各个步骤，但是，这并非要求或者暗示必须按照该特定顺序来执行这些步骤，或是必须执行全部所示的步骤才能实现期望的结果。附加的或备选的，可以省略某些步骤，将多个步骤合并为一个步骤执行，以及/或者将一个步骤分解为多个步骤执行等。
通过以上的实施方式的描述，本领域的技术人员易于理解，这里描述的示例实施方式可以通过软件实现，也可以通过软件结合必要的硬件的方式来实现。因此，根据本公开实施方式的技术方案可以以软件产品的形式体现出来，该软件产品可以存储在一个非易失性存储介质(可以是CD-ROM，U盘，移动硬盘等)中或网络上，包括若干指令以使得一台计算设备(可以是个人计算机、服务器、移动终端、或者网络设备等)执行根据本公开实施方式的方法。
本领域技术人员在考虑说明书及实践这里公开的发明后，将容易想到本公开的其它实施方案。本公开旨在涵盖本公开的任何变型、用途或者适应性变化，这些变型、用途或者适应性变化遵循本公开的一般性原理并包括本公开未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的，本公开的真正范围和精神由所附的权利要求指出。

Claims (14)

  1. 一种瞳孔检测方法，其中，包括：
    检测被检测对象的位置信息，在根据所述位置信息确定所述被检测对象进入目标区域内后，获取所述被检测对象的多幅场景图像；
    对所述多幅场景图像进行人脸识别处理，得到多幅人脸图像；
    检测所述多幅人脸图像中的人眼，得到多幅人眼图像；
    检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息。
  2. 根据权利要求1所述的方法，其中，所述方法还包括：
    在检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息之后，对所述多幅人脸图像进行去重处理，针对一个所述被检测对象保留一幅人脸图像；
    确定保留的所述人脸图像中的瞳孔的信息为所述被检测对象的瞳孔检测结果。
  3. 根据权利要求2所述的方法，其中，所述方法还包括：
    在确定保留的所述人脸图像中的瞳孔的信息为所述被检测对象的瞳孔检测结果之后，输出所述被检测对象的瞳孔的信息以及所述被检测对象的人脸图像。
  4. 根据权利要求1所述的方法，其中，所述方法还包括：
    在检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息之前，对所述多幅人眼图像进行质量评估；
    检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息，包括：
    检测所述多幅人眼图像中通过质量评估的人眼图像中的瞳孔，得到通过质量评估的人眼图像中的瞳孔的信息。
  5. 根据权利要求1所述的方法，其中，获取所述被检测对象的多幅场景图像，包括：
    在所述目标区域有被检测对象出现时，获取所述目标区域内所述被检测对象的位置信息；
    若根据获取的所述被检测对象的位置信息确定所述被检测对象在移动，降低用以获取所述场景图像的图像采集装置的曝光时间至预设曝光时间，并通过所述图像采集装置对所述目标区域进行拍摄，得到多幅所述目标区域的场景图，其中，所述预设曝光时间不长于第一预设时长。
  6. 根据权利要求5所述的方法，其中，所述方法还包括：
    在通过所述图像采集装置对所述目标区域进行拍摄时，对所述目标区域进行补光。
  7. 根据权利要求1所述的方法，其中，所述目标区域为距离拍摄所述多幅场景图像的图像采集装置不超过所述图像采集装置的景深范围的区域。
  8. 根据权利要求1所述的方法，其中，检测被检测对象的位置信息，包括：
    通过测距传感器检测所述被检测对象的位置信息，其中，所述测距传感器与所述图像采集装置位于同一竖直面上。
  9. 根据权利要求8所述的方法，其中，所述图像采集装置至少包括：
    在所述竖直面上纵向分布的至少两个图像采集装置，所述至少两个图像采集装置的成像区域重叠。
  10. 根据权利要求2所述的方法，其中，对所述多幅人脸图像进行去重处理，针对一个所述被检测对象保留一幅人脸图像，包括：
    依次获取所述多幅人脸图像，在每次获取到所述人脸图像后，对比当前人脸图像与上一幅人脸图像的图像质量以及人脸相似度，若当前人脸图像与上一幅人脸图像之间的人脸相似度高于相似度阈值，保留当前人脸图像与上一幅人脸图像中图像质量高的人脸图像，直至获取的人脸图像达预设数量后输出保留的人脸图像，若当前人脸图 像与上一幅人脸图像相似度不高于相似度阈值或当前人脸图像的拍摄时间与上一幅人脸图像的拍摄时间之差不小第二预设时长，将所述当前人脸图像作为所述被检测对象的人脸图像进行输出。
  11. 一种瞳孔检测装置，其中，包括：
    获取模块，被配置为检测被检测对象的位置信息，在根据所述位置信息确定所述被检测对象进入目标区域内，获取所述被检测对象的多幅场景图像；
    人脸识别模块，被配置为对所述多幅场景图像进行人脸识别处理，得到多幅人脸图像；
    人眼检测模块，被配置为检测所述多幅人脸图像中的人眼，得到多幅人眼图像；
    瞳孔检测模块，被配置为检测所述多幅人眼图像中的瞳孔，得到所述多幅人眼图像的瞳孔的信息。
  12. 一种电子设备，其中，包括：
    处理器；以及
    存储器，用于存储所述处理器的可执行指令；
    其中，所述处理器配置为经由执行所述可执行指令来执行权利要求1至10中任一项所述的瞳孔检测方法。
  13. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现权利要求1至10中任一项所述的瞳孔检测方法。
  14. 一种计算机程序产品，包括计算机程序，其中，所述计算机程序被处理器执行时实现权利要求1-10中任一项所述的方法的步骤。
PCT/CN2023/078539 2022-06-01 2023-02-27 瞳孔检测方法、装置、存储介质及电子设备 WO2023231479A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210623020.9 2022-06-01
CN202210623020.9A CN114926891A (zh) 2022-06-01 2022-06-01 瞳孔检测方法、装置、存储介质及电子设备

Publications (1)

Publication Number Publication Date
WO2023231479A1 true WO2023231479A1 (zh) 2023-12-07

Family

ID=82812755

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/078539 WO2023231479A1 (zh) 2022-06-01 2023-02-27 瞳孔检测方法、装置、存储介质及电子设备

Country Status (2)

Country Link
CN (1) CN114926891A (zh)
WO (1) WO2023231479A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114926891A (zh) * 2022-06-01 2022-08-19 同方威视科技江苏有限公司 瞳孔检测方法、装置、存储介质及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109858337A (zh) * 2018-12-21 2019-06-07 普联技术有限公司 一种基于瞳孔信息的人脸识别方法、系统和设备
WO2020038069A1 (zh) * 2018-08-22 2020-02-27 Oppo广东移动通信有限公司 曝光控制方法、装置和电子设备
CN111738241A (zh) * 2020-08-20 2020-10-02 宁波均联智行科技有限公司 基于双摄像头的瞳孔检测方法及装置
CN112800815A (zh) * 2019-11-13 2021-05-14 北京卡尔斯通科技有限公司 一种基于深度学习的视线方向估计方法
CN113205057A (zh) * 2021-05-13 2021-08-03 北京百度网讯科技有限公司 人脸活体检测方法、装置、设备及存储介质
CN114926891A (zh) * 2022-06-01 2022-08-19 同方威视科技江苏有限公司 瞳孔检测方法、装置、存储介质及电子设备

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020038069A1 (zh) * 2018-08-22 2020-02-27 Oppo广东移动通信有限公司 曝光控制方法、装置和电子设备
CN109858337A (zh) * 2018-12-21 2019-06-07 普联技术有限公司 一种基于瞳孔信息的人脸识别方法、系统和设备
CN112800815A (zh) * 2019-11-13 2021-05-14 北京卡尔斯通科技有限公司 一种基于深度学习的视线方向估计方法
CN111738241A (zh) * 2020-08-20 2020-10-02 宁波均联智行科技有限公司 基于双摄像头的瞳孔检测方法及装置
CN113205057A (zh) * 2021-05-13 2021-08-03 北京百度网讯科技有限公司 人脸活体检测方法、装置、设备及存储介质
CN114926891A (zh) * 2022-06-01 2022-08-19 同方威视科技江苏有限公司 瞳孔检测方法、装置、存储介质及电子设备

Also Published As

Publication number Publication date
CN114926891A (zh) 2022-08-19

Similar Documents

Publication Publication Date Title
US10949952B2 (en) Performing detail enhancement on a target in a denoised image
US11182592B2 (en) Target object recognition method and apparatus, storage medium, and electronic device
CN108090458B (zh) 人体跌倒检测方法和装置
US8314854B2 (en) Apparatus and method for image recognition of facial areas in photographic images from a digital camera
JP6555906B2 (ja) 情報処理装置、情報処理方法、およびプログラム
CN109711318B (zh) 一种基于视频流的多人脸检测与跟踪方法
CN104811660A (zh) 控制装置及控制方法
WO2019011073A1 (zh) 人脸活体检测方法及相关产品
JP2019186955A (ja) 情報処理システム、情報処理方法及びプログラム
CN108513110A (zh) 人脸识别监控摄像头
CN111444555B (zh) 一种测温信息显示方法、装置及终端设备
WO2023231479A1 (zh) 瞳孔检测方法、装置、存储介质及电子设备
CN108540707A (zh) 人脸识别执法仪
CN108259769B (zh) 图像处理方法、装置、存储介质及电子设备
JP6991045B2 (ja) 画像処理装置、画像処理装置の制御方法
US20240048672A1 (en) Adjustment of shutter value of surveillance camera via ai-based object recognition
US11132778B2 (en) Image analysis apparatus, image analysis method, and recording medium
CN108960146A (zh) 人脸识别监控视频的图像预处理方法
CN112949467A (zh) 人脸检测方法、装置、电子设备和存储介质
CN107667522B (zh) 形成活动图像的方法和装置
CN109040604B (zh) 拍摄图像的处理方法、装置、存储介质及移动终端
CN109218620B (zh) 基于环境亮度的拍照方法、装置、存储介质及移动终端
JP5128454B2 (ja) 瞼検出装置、瞼検出方法及びプログラム
CN116071813A (zh) 一种场景检测方法、装置、介质及设备
JP5995610B2 (ja) 被写体認識装置及びその制御方法、撮像装置、表示装置、並びにプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23814678

Country of ref document: EP

Kind code of ref document: A1