WO2017143679A1 - 一种辐射源控制方法和速通式安检系统 - Google Patents

一种辐射源控制方法和速通式安检系统 Download PDF

Info

Publication number
WO2017143679A1
WO2017143679A1 PCT/CN2016/083619 CN2016083619W WO2017143679A1 WO 2017143679 A1 WO2017143679 A1 WO 2017143679A1 CN 2016083619 W CN2016083619 W CN 2016083619W WO 2017143679 A1 WO2017143679 A1 WO 2017143679A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
radiation source
vehicle
detected object
template
Prior art date
Application number
PCT/CN2016/083619
Other languages
English (en)
French (fr)
Inventor
李苏祺
曹艳锋
王少锋
郑建斌
胡晓伟
闫雄
凌敏
Original Assignee
北京君和信达科技有限公司
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 北京君和信达科技有限公司 filed Critical 北京君和信达科技有限公司
Publication of WO2017143679A1 publication Critical patent/WO2017143679A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V5/00Prospecting or detecting by the use of ionising radiation, e.g. of natural or induced radioactivity
    • G01V5/20Detecting prohibited goods, e.g. weapons, explosives, hazardous substances, contraband or smuggled objects
    • G01V5/22Active interrogation, i.e. by irradiating objects or goods using external radiation sources, e.g. using gamma rays or cosmic rays
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01VGEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V5/00Prospecting or detecting by the use of ionising radiation, e.g. of natural or induced radioactivity

Definitions

  • the invention relates to the technical field of radiation imaging, in particular to a radiation source control method and a quick-flow security inspection system using the same.
  • the non-stop inspection technology based on the automatic scanning process control of the radiation source has a very high throughput rate and high safety inspection efficiency.
  • the non-stop inspection technology in order to ensure the safety of the driver in the cab, it is necessary to evade the cab, that is, the high-energy, high-dose-rate radiation beam is allowed to be emitted only after the cab leaves the radiation inspection position, only to the rear of the cab. Cargo cargo is scanned for inspection. There is a security hole because the cab is not scanned.
  • the vehicle-specific scanning technology can be used to check the cab portion of the vehicle with low-energy or low-dose radiation to meet the radiation safety standard, and the cargo compartment is inspected with high-energy or high-dose radiation.
  • the premise of this non-stop inspection technology is that it is necessary to accurately determine the position of the cab, and the accuracy of the judgment directly affects the safety of the personnel in the cab.
  • the present invention provides a radiation source control method and a rapid-flow security inspection system, which performs real-time processing on a scanned image of each ray pulse, and determines whether the protected area (or the entire vehicle) of the vehicle has passed through image analysis. Scan location.
  • the signal I s corresponds.
  • Embodiments of the present invention identify the boundary image of the cab and the cargo area for the image of the vehicle acquired by the radiation imaging, thereby controlling the radiation source to emit suitable rays for the purpose of vehicle inspection, and the entire process is not affected by environmental factors, nor It will be affected by the boundary position cover, which can minimize the number of sensors used, reduce installation and maintenance costs, and have high detection reliability.
  • 1-3 are respectively a grayscale diagram, a binary image, and related calculation values of three models of the embodiment of the present invention.
  • FIG. 4 is a flow chart of a radiation source control method according to an embodiment of the present invention.
  • 5-7 are flowcharts showing three processing procedures of the image recognition algorithm in the embodiment of the present invention, respectively.
  • Figure 8 is a flow diagram of controlling a radiation source to stop emitting a radiation beam in accordance with an embodiment of the present invention.
  • the cab of a cargo vehicle and other areas that can accommodate passengers are usually a whole, called the front, and the cargo area behind the front is the cargo compartment, and there is a gap between the front and the cargo compartment.
  • the width of this gap may vary for different types of trucks. As shown in Fig. 1, the width of the gap between the front and the cargo compartment of the truck is large, and the gap width of Fig. 2 is small.
  • the front and rear seats of the passenger car can accommodate passengers, so the passenger vehicle itself is a whole, and there is no gap on the body, as shown in Figure 3.
  • the basic idea of the embodiment of the present invention is to determine whether the protected area of the inspected vehicle (the truck head or the passenger vehicle as a whole) has passed the scanning area by detecting the integrity of the scanned object by the radiation image. According to the characteristics of the ray scanning, the scanned image (each frame of each pulse) can be processed in real time, and the object to be inspected is extracted from the background by the image segmentation method.
  • the appearance of the front of the vehicle is first detected as The beginning of the whole protection area, the frame-by-frame image is detected, and whether the target is completely finished according to whether the area of the scanned object in each frame suddenly decreases, and when the whole is over, it is determined that the vehicle protection area has passed the scanning, and then the The rays are adjusted from low-energy rays or low-dose-rate rays to high-energy rays or high-dose-rate rays, and the cargo compartment behind the front of the vehicle is scanned.
  • a radiation source control method includes the following steps:
  • the image recognition algorithm detects whether the current scanning position is a boundary position of the detected object, wherein the boundary position refers to a gap between the first portion and the second portion of the detected object, the current scanning position and the sth detector signal I s corresponding;
  • the control radiation source starts to emit the second radiation beam.
  • the above solution is applied to the non-stop inspection technology of the vehicle, the vehicle is the detected object, the vehicle travels into the inspection channel, the vehicle moves relative to the radiation source, and the front end of the detection vehicle reaches the upstream side of the scanning position of the radiation source.
  • the radiation source is controlled to emit the first radiation beam, and the first part of the vehicle (the passenger area of the vehicle head) is scanned and inspected.
  • the recognition algorithm detects that the current scanning position of the radiation source (corresponding to the sth detector signal I s ) is located at a boundary position between the first portion and the second portion (the cargo area behind the vehicle head), and the control radiation source emits the second radiation beam, Scanning inspection of the cargo area behind the passenger area of the vehicle.
  • the first type of radiation beam refers to a low energy or low dose rate radiation beam that meets the radiation safety standard
  • the second type of radiation beam refers to a high energy or high dose rate radiation beam.
  • the non-stop scanning inspection of the vehicle can be performed by using the technical solution of the embodiment of the present invention, and the image recognition algorithm is used to identify the position of the front head during the scanning process, and the radiation source scans and inspects the front end of the vehicle with low energy or low dose rate when the front end reaches the scanning position. After scanning the position and the cargo compartment is about to reach the scanning position, the radiation source is switched to emit a high energy or high dose rate ray scan to check the cargo compartment cargo. It not only realizes the differential scanning of the whole vehicle, but also has high security inspection efficiency and low miss detection rate. It also eliminates the trouble of installing special sensor equipment for identifying the front position of the vehicle, reduces the input cost, and is not affected by environmental factors such as occlusion and rain and snow. Identification is accurate and stable.
  • the image recognition algorithm may be implemented by different methods.
  • the embodiment of the present invention determines the protection area of the inspected vehicle by detecting the integrity of the scanned object by the radiation image (the front of the truck) Or whether the passenger vehicle as a whole has passed the scanning area. Therefore, according to the basic principle of the ray scanning, the scanned image of each frame of the detector can be processed in real time, and the object to be inspected is extracted from the background by the image segmentation method. First, the presence of the front of the vehicle is detected.
  • the frame-by-frame image is detected, and whether the target protection area is completely terminated according to whether the area of the scanned object in each frame suddenly decreases, when the whole ends, It is determined that the vehicle protection area has been scanned, and the radiation can be adjusted from low energy or low dose rate to high energy or high dose rate, and the cargo compartment behind the vehicle head is scanned.
  • Figure 5-7 shows the processing flow of three image recognition algorithms, respectively, which are described in detail below.
  • the radiation source of such equipment is a pulse working mode, in which the secondary detector acquisition time corresponds to each ray pulse time, and each frame of the detector is a ray scan image of each pulse.
  • FIG. 1(b), 2(b) and 3(b) is the binary map of the three models, and Ib i is the value of the ith column in the diagram.
  • Figures 1(a), 2(a) and 3(a) are grayscale diagrams for three models.
  • the threshold value may be set to the gray value of the unloaded portion (excluding the object to be inspected) in the corrected image multiplied by a proportional coefficient, that is, ⁇ *I CAir , where I CAir is the gray scale of the hollow portion of the corrected image I Ci
  • is a proportional coefficient, 0 ⁇ ⁇ ⁇ 1; and, pixels smaller than ⁇ *I CAir are set to 0, and pixels larger than or equal to ⁇ *I CAir are set to 1.
  • the scale factor can be adjusted according to the recognition effect, and its value range is ⁇ 1, for example, it can be set to 0.85.
  • the projection value is the number of non-background value pixels, and the projection value P i can be obtained.
  • the typical vehicle image projection value is shown in Figure 1-3;
  • the difference window can be set larger, for example, set to 10, and the difference formula is as follows:
  • G i (P i-1 +P i-2 +...+P iw/2 -P iw/2-1 -P iw/2-2 -...- P iw )/w,
  • the moving window detection method can also directly measure the change of the projection value to realize whether the vehicle protection area has passed the detection of the scanning position, and the flow chart is shown in FIG. 6 . Proceed as follows:
  • the purpose of selecting the window to calculate the mean value of the projection value is to perform signal smoothing processing to reduce the interference of small objects, and the window width may be appropriately larger, for example, the width is 9, and the specific effect needs to be set according to the implementation effect;
  • step 3 calculate the current projection value mean in the CurrentP window as follows:
  • the width of the CurrentP window may be appropriately smaller, for example, the width is 3, which is specifically set according to the smoothness required in the implementation process;
  • Thrd_P is a discriminating threshold coefficient, and its value range is ⁇ 1, for example, it can be set to 0.3, and can be specifically set according to the implementation effect.
  • Algorithm 3 according to the real-time scanned image, in addition to the above two uses image statistical features to achieve In addition to the method of detecting the protection area of the vehicle, the pattern recognition technology may be used to identify the current scanning vehicle to determine whether the protected area of the vehicle passes the scanning position, and the algorithm 3 is to implement detection by template matching.
  • the main idea of the algorithm is to first establish a template library, and the protection areas of different types of vehicles (truck front part or passenger vehicle) correspond to different templates; when the inspected vehicle scans, the scanned image obtained in real time is treated as The object is matched with each template in the template library; when the matching degree of the scanned image with a certain template reaches the corresponding requirement, the identification is completed, and it is determined that the protected area of the vehicle has passed the scanning position.
  • Specific steps are as follows:
  • the first type of radiation beam is used to scan the common types of vehicles, after obtaining the original scanned image, the inconsistency correction, the brightness correction (see step 3 of Algorithm 1), and the geometric correction, etc.
  • the geometric correction can be based on the vehicle and
  • the relative speed between the scanning systems is processed to eliminate image distortion caused by different speeds, so that the processed images are at the same measurement scale, so that the current detection image and the template sample are prevented from being unrecognizable due to different ratios.
  • the matching process is dynamically performed in real time with the scan.
  • the obtained image can be matched once in each ray pulse period, or several pulse periods can be selected for matching, which can be comprehensively measured according to the matching operation speed and the pulse frequency.
  • the matching period is determined under the premise of ensuring real-time performance.
  • the specific matching process is shown in Figure 7, and the steps are as follows:
  • r is equal to 0, the process proceeds to step 5), and the matching process is performed; when r is not equal to 0, the process proceeds to step 2).
  • the P value can be determined according to the matching operation speed and the pulse frequency, so as to ensure that the real-time P should be as small as possible;
  • Thrd_w is the width control threshold, which can be set according to the specific implementation effect, generally a few pixels width, for example, 5.
  • the degree of matching can be measured by calculating the degree of non-similarity, and a more efficient matching strategy, such as a sequential similarity detection algorithm, can be used in implementation ( Sequential Similarity Detection Algorithm (SSDA), the calculation formula of non-similarity is as follows:
  • the current scanned image is considered to match the template, that is, the protected area of the vehicle (the front or passenger car) has been detected to have passed the scanning position.
  • the ray can be adjusted from the first radiation beam to the second radiation beam, and the process ends; otherwise, return to step 2.
  • the above embodiments describe in detail the use of the present invention to control the switching of the radiation beam (from the first The first type of radiation beam checked by the part of the passenger area is switched to the second type of radiation beam used for the second part of the cargo area inspection, and three image recognition algorithms are given.
  • the following embodiments describe the use of the present invention to control the stop of the radiation beam, i.e., to control the stop of the radiation beam as it exits the scanning position (scanning complete).
  • the steps of the control method for stopping the radiation beam of the radiation source are as follows:
  • the radiation source is controlled to stop emitting the radiation beam.

Landscapes

  • Physics & Mathematics (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Geophysics (AREA)
  • Image Processing (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)
  • Image Analysis (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

一种辐射源控制方法,包括:在辐射源开始发射第一辐射束扫描被检测物之后,通过辐射束探测器获取探测器信号;通过图像识别算法对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置;当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射第二辐射束。该方法能够准确快捷地判断车辆驾驶室和货物区域的分界位置。还公开了一种速通式安检系统。

Description

一种辐射源控制方法和速通式安检系统 技术领域
本发明涉及辐射成像技术领域,具体涉及一种辐射源控制方法和利用此方法的速通式安检系统。
背景技术
在车辆安全检查过程中,基于辐射源自动扫描流程控制的不停车检查技术具有非常高的通过率,安全检查效率高。在不停车检查技术中,为保证驾驶室内司机的安全,需要对驾驶室进行辐射避让,即只有当驾驶室离开辐射检查位置之后才允许发射高能、高剂量率辐射束,仅对驾驶室后边的货厢货物进行扫描检查。由于不对驾驶室做扫描检查,因此存在安全漏洞。为消除该安全漏洞,可采用全车区别扫描技术,以满足辐射安全标准的低能或低剂量率辐射对车辆驾驶室部分进行检查,而采用高能或高剂量率辐射对货厢部分进行检查。这种不停车检查技术的前提是,需要准确判断驾驶室的位置,判断的准确程度直接影响驾驶室内人员的安全。
目前用于车辆不停车检查的辐射源控制方案大多利用地感线圈、光幕、光电传感器、激光距离传感器、激光扫描仪等硬件设备来识别驾驶室位置,可在一定程度上满足使用需求,但是还存在如下弊端:一方面,传感器等硬件设备需要进行专门安装和布置,前期投入成本高,并且各种设备容易受到环境的影响(如雨雪、风沙等),设备故障率高,维护成本高;另一方面,传感器等硬件设备容易受车辆装饰物、覆盖物等的影响而发生识别错误,导致货厢漏扫描。
发明内容
有鉴于此,本发明提出一种辐射源控制方法和速通式安检系统,对每个射线脉冲的扫描图像进行实时处理,通过图像分析的方法确定车辆的保护区域(或整车)是否已经通过扫描位置。
一方面,本发明提供一种辐射源控制方法,包括:在辐射源开始发射第一辐射束扫描被检测物之后,通过辐射束探测器获取探测器信号Ii,其中i为探测器信号的顺序号,i=0,1,2,…,n;通过图像识别算法对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置;当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射第二辐射束;其中,所述分界位置是指被检测物的第一部分与第二部分之间的空隙,所述当前扫描位置与第s个探测器信号Is相对应。
另一方面,本发明还提供一种速通式安检系统,包括:至少一个辐射源,用于发射至少两种辐射束;辐射束探测器,用于获取多个探测器信号Ii,其中i为探测器信号的顺序号,i=0,1,2,…,n;图像识别算法模块,用于对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置;辐射源控制模块,用于当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射指定辐射束;其中,所述分界位置是指被检测物的第一部分与第二部分之间的空隙,所述当前扫描位置与第s个探测器信号Is相对应。
本发明的实施例对辐射成像获取的车辆图像进行驾驶室和货物区域分界位置的识别,进而控制辐射源发出适合的射线,达到车辆检查的目的,整个过程不会受到环境因素的影响,也不会受到分界位置覆盖物的影响,能够最大限度地减少传感器的使用数量,降低了安装维护成本,检测可靠性高。
附图说明
图1-3分别是本发明实施例三种车型的灰度图、二值图及相关计算值。
图4是本发明实施例的辐射源控制方法流程图。
图5-7分别是本发明实施例中图像识别算法的三种处理过程的流程图。
图8是本发明实施例控制辐射源停止发射辐射束的流程图。
具体实施方式
以下结合附图以及具体实施例,对本发明的技术方案进行详细描述。
载货车辆的驾驶室以及其他可以容纳乘客的区域通常是一个整体,称为车头,车头后方的载货区域为货厢,车头与货厢之间存在间隙。对于不同型号的货车,这个间隙的宽度可能不同。如图1货车的车头与货厢之间的间隙宽度较大,图2的间隙宽度较小。与货车不同,载客车辆的车厢前后排座位都可以容纳乘客,因此载客车辆本身是一个整体,车身上不存在间隙,如图3。
本发明实施例的基本思想是,通过辐射图像检测被扫描物体的整体性,来判断被检车辆的保护区域(货车车头或载客车辆整体)是否已经经过扫描区域。根据射线扫描的特点,可对每个脉冲的扫描图像(每帧图像)进行实时处理,通过图像分割的方法从背景中提取出被检查目标,在此基础上,首先检测到车头的出现,作为保护区域整体的开始,逐帧图像进行检测,根据每帧图像中被扫目标的面积是否突然减少,来判断该目标整体是否结束,当整体结束时,判定车辆保护区域已经通过扫描,进而可将射线从低能射线或低剂量率射线调整为高能射线或高剂量率射线,对车头后方的货厢进行扫描检查。
参考图4,本发明实施例的辐射源控制方法包括以下步骤:
在辐射源开始发射第一辐射束扫描被检测物之后,通过辐射束探测器获取探测器信号Ii,其中i为探测器信号的顺序号,i=0,1,2,…,n;
通过图像识别算法检测当前扫描位置是否为被检测物的分界位置,其中所述分界位置是指被检测物的第一部分与第二部分之间的空隙,当前扫描位置与第s个探测器信号Is相对应;
当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射第二辐射束。
在实际应用场景中,将上述方案应用于车辆的不停车检查技术中,车辆即为被检测物,车辆行驶进入检查通道,车辆相对辐射源移动,检测车辆前端到达辐射源扫描位置的上游侧的预定位置的时刻,当检测到车辆前端到达该预定位置时(说明车头即将进入辐射源的扫描位置),控制辐射源发射第一种辐射束,对车辆第一部分(车头载客区域)进行扫描检查;在扫描过程中,采用辐射探测器按时间顺序以一定的频率实 时获取辐射束信号Ii,其中i为探测器信号的顺序号(i=0,1,2,…,n);通过图像识别算法检测到辐射源当前扫描位置(对应第s个探测器信号Is)位于第一部分和第二部分(车头后方的载货区域)的分界位置时,控制辐射源发射第二种辐射束,对车辆载客区域后边的载货区域进行扫描检查。其中,第一种辐射束是指符合辐射安全标准的低能或低剂量率辐射束,第二种辐射束是指高能或高剂量率辐射束。
利用本发明实施例的技术方案可进行车辆的不停车扫描检查,在扫描过程中利用图像识别算法识别车头位置,在车头到达扫描位置时辐射源以低能或低剂量率射线扫描检查车头,当车头通过扫描位置之后且货厢即将到达扫描位置时,将辐射源切换为发射高能或高剂量率射线扫描检查货厢货物。既实现了全车区别扫描,安检效率高,漏检率低,又免除了以往需要安装识别车头位置的专用传感器设备的麻烦,降低了投入成本,并且不受遮挡和雨雪等环境因素影响,识别准确、稳定。
在本发明的实施例中,其中的图像识别算法可采用不同的方法实现,举例来说,本发明实施例是通过辐射图像检测被扫描物体的整体性来判断被检车辆的保护区域(货车车头或载客车辆整体)是否已经经过扫描区域,因此,根据射线扫描基本原理,可对探测器每帧扫描图像进行实时处理,通过图像分割的方法从背景中提取出被检查目标,在此基础上,首先检测到车头的出现,作为保护区域整体的开始,逐帧图像进行检测,根据每帧图像中被扫目标的面积是否突然减少,来判断该目标保护区域整体是否结束,当整体结束时,判定车辆保护区域已经通过扫描,进而可将射线从低能或低剂量率调整为高能或高剂量率,对车头后方的货厢进行扫描检查。
图5-7分别给出三种图像识别算法的处理流程图,以下分别详细描述。通常此类设备的射线源都是脉冲工作模式,此时次探测器采集时间与每个射线脉冲时间对应,探测器每帧图像即为每个脉冲的射线扫描图像。
算法一,参考图5,步骤如下:
1)车辆进入扫描区域,打开第一种辐射束开关,射线源发出第一种辐射束进行扫描,初始化计数器Ctr=0;
2)线阵探测器采集每个射线脉冲的数据,获得每列扫描图像Ii,其中i=0,1,…,N,N为脉冲计数;
3)对图像Ii每个像素作不一致性校正和亮度校正,得到校正图像Ici;其中不一致性校正和亮度校正可按已知校正方法处理;
4)采用阈值法对图像Ici进行二值化处理,实现背景和被扫目标的图像分割,得到二值化图像Ibi,二值化效果可参考图1(b)、2(b)和3(b),分别是三种车型的二值图,Ibi即为示意图中第i列值。图1(a)、2(a)和3(a)为三种车型的灰度图。阈值可设置为校正后图像中的空载部分(不含被检物)的灰度值乘以比例系数,即μ*ICAir,其中,ICAir为校正图像ICi中空载部分的灰度值,μ为比例系数,0<μ<1;并且,将小于μ*ICAir的像素设为0,将大于等于μ*ICAir的像素设为1。比例系数可根据识别效果进行调整,其取值范围<1,例如可设为0.85。
5)对二值图像进行垂直投影计算,投影值即为非背景值像素的数量,可得到投影值Pi,典型车辆图像投影值如图1-3中所示;
6)计算投影值的梯度,可用差分法进行计算,例如中心差分,为提高对投影值缓慢变化的检测效果,可将差分窗口设大一些,例如设为10,差分公式如下:
Gi=(Pi-1+Pi-2+…+Pi-w/2-Pi-w/2-1-Pi-w/2-2-…-Pi-w)/w,
其中w为窗口宽度。每个射线脉冲图像投影值和梯度值如图1(c)、2(c)和3(c)所示。
7)当Ctr=0时,转入步骤8;当Ctr=1时,转入步骤9;当Ctr不等于0且不等于1时,转入步骤10);
8)当投影值梯度大于设定的阈值Thrd_Gf时(Gi>Thrd_Gf,如图1-3中所示),检测到车头,则将Ctr设为1,即Ctr=1,返回步骤2);否则直接返回步骤2);
9)当投影值梯度小于设定的阈值Thrd_Gb时(Gi<Thrd_Gb,如图1-3中所示),检测到驾驶室后沿开始下降,将Ctr设为2,即Ctr=2,返回步骤2);否则直接返回步骤2);
10)当当前投影值梯度Gi大于上一梯度值Gi-1与阈值Thrd_IP之和时 (Gi>Gi-1+Thrd_IP),即检测到梯度变化的拐点,如图1-3中所示,则可确定已扫描至保护区域后部的间隙部分,说明车辆的保护区域已经通过扫描,进而可将射线从第一种辐射束调整为第二种辐射束;否则,返回步骤2。
算法二,根据实时的扫描图像,利用图像分割的方法,还可以通过移动窗口检测的方式直接衡量投影值的变化,实现车辆保护区域是否已经通过扫描位置的检测,流程图如图6所示,步骤如下:
1)车辆进入扫描区域,打开第一种辐射束开关,射线源发出第一种辐射束进行扫描,初始化投影最大值Pmax=0;
2)同前述算法一中的步骤2)-5);
3)按下式计算MaxP窗口中的投影值均值:
Mi=(Pi+Pi-1+…+Pi-m+1)/m,其中Mi为当前投影值前m列投影值的均值,m为MaxP窗口的宽度,m≥1,当m=1时,Mi=Pi
其中,选取窗口计算投影值均值的目的是进行信号平滑处理,减小细小物体的干扰,窗口宽度可适当取大一些,例如宽度为9,具体需按实施效果进行设置;
4)当Mi>Pmax时,更新Pmax值,即Pmax=Mi
5)采用步骤3)中类似的方法,按下式在CurrentP窗口中计算当前投影值均值:
Ci=(Pi+Pi-1+…+Pi-c+1)/c,其中Ci为当前投影值前c列投影值的均值,c为CurrentP窗口的宽度,其中c≥1,当c=1时,Ci=Pi
其中,CurrentP窗口的宽度可适当取小一些,例如宽度为3,具体根据实施过程中需要的平滑程度进行设置;
6)当Ci<Pmax*Thrd_P时,可确定已扫描至保护区域后部的间隙部分,说明车辆的保护区域已经通过扫描,进而可将射线从第一种辐射束调整为第二种辐射束;否则,返回步骤2;
其中Thrd_P为判别阈值系数,其取值范围<1,例如可设为0.3,具体可按实施效果择优设置。
算法三,根据实时扫描图像,除了上述两种利用图像统计特征实现 车辆的保护区域检测的方式外,还可以采用模式识别技术,对当前扫描车辆进行识别,以判断车辆的保护区域是否通过扫描位置,算法三即为采用模板匹配的方式实现检测。
本算法的主要思想是:先建立模板库,不同类型车辆的保护区域(货车车头部分或载客车辆整车)对应不同的模板;当被检车辆进行扫描时,将实时获得的扫描图像作为处理对象,与模板库中的各模板进行匹配;当扫描图像与某一模板匹配度达到相应要求时,则完成识别,同时确定车辆的保护区域已经通过扫描位置。具体步骤如下:
1.建库
首先,用第一种辐射束对常见类型车辆进行扫描,获得原始扫描图像后,进行不一致性校正、亮度校正(参见算法一的步骤3))以及几何校正等预处理,几何校正可根据车辆与扫描系统之间的相对速度进行处理,消除不同速度引起的图像畸变,使处理后图像处于同一测量尺度,这样可避免当前检测图像与模板样本因比例不同而导致无法识别。
在各预处理后的图像中人工选取相应的车辆保护区域(货车车头部分或载客车辆整车),作为不同模板Tn,其中n=1,2,…,N,N为模板个数。
2.匹配
匹配的过程是随扫描实时动态进行的,可以在每个射线脉冲周期对已获得的图像进行一次匹配,也可以选择几个脉冲周期进行一次匹配,具体可根据匹配运算速度以及脉冲频率快慢综合衡量,在保证实时性的前提下确定匹配周期。每次匹配前,根据当前扫描图像的宽度从模板库Tn(只是货厢以前部分,包含缝隙)中选取宽度相当的模板组成有效模板组VTm,匹配则在有效模板组中进行,这样能减少匹配的运算量,提高识别效率。具体匹配过程如图7所示,步骤如下:
1)车辆进入扫描区域,打开第一种辐射束开关,射线源发出第一种辐射束进行扫描;
2)同前述算法一中的步骤2)、3)对探测器信号Ii进行校正, 得到校正信号ICi
3)将已经获取的前i列校正信号IC1~ICi按先后顺序进行组合,并根据车辆移动速度进行几何校正,得到当前时刻已经完成的车辆扫描图像Isumi
4)判断当前脉冲是否进行匹配处理,设r=i%p,其中,i为当前脉冲序号,p为设置的匹配间隔脉冲数(每隔p个脉冲进行一次匹配),%为取余运算。当r等于0时,转入步骤5),进行匹配处理;当r不等于0时,转入步骤2)。P值可根据匹配运算速度以及脉冲频率快慢综合衡量确定,为了保证实时性P应尽可能小;
5)根据图像Isumi的宽度,按下式从模板库Tn中选择宽度接近的模板,组成有效模板组VTm
VTm∈Tn,m=1,2,3,…,
Figure PCTCN2016083619-appb-000001
其中,m为有效模板组中模板数量,
Figure PCTCN2016083619-appb-000002
为图像Isumi的宽度,
Figure PCTCN2016083619-appb-000003
为图像VTm的宽度,Thrd_w为宽度控制阈值,可根据具体实施效果进行设置,一般为几个像素宽度即可,例如5。
6)将图像Isumi与有效模板组VTm中的每个模板进行匹配,可通过计算非相似度来衡量匹配程度,实施时可采用效率较高的匹配策略,例如贯序相似性检测算法(Sequential Similarity Detection Algorithm,SSDA),非相似度计算公式如下:
Figure PCTCN2016083619-appb-000004
其中,
Figure PCTCN2016083619-appb-000005
分别为图像Isumi的宽度和高度;
7)当非相似度小于设定的阈值Thrd_S,即Sm(u)<Thrd_S时,则认为当前扫描图像与模板匹配,即检测到车辆的保护区域(车头或载客汽车)已经通过扫描位置,进而可将射线从第一种辐射束调整为第二种辐射束,流程结束;否则,返回步骤2。
以上实施例详细描述了利用本发明控制辐射束的切换(从用于第一 部分载客区域检查的第一种辐射束切换到用于第二部分载货区域检查的第二种辐射束),并给出三种图像识别算法。以下实施例描述利用本发明控制辐射束停止,即车辆尾部离开扫描位置(扫描完成)时控制停止辐射束。
参考图8,辐射源辐射束停止的控制方法步骤如下:
采用辐射源发射的辐射束对被检测物体进行辐射扫描检查;
采用阵列辐射探测器按时间顺序以一定的频率获取辐射束信号Ii,其中i为探测器信号的顺序号(i=0,1,2,…);
当通过图像识别算法识别出当前扫描位置对应空气且被检测物位于扫描位置下游的特征(即被检物已经通过扫描),控制辐射源停止发射辐射束。
在上述几个实施例中,通过辐射图像实时检测的方式,除了可以判别出车辆的保护区域(车头或载客汽车)是否已经经过扫描,作为辐射束切换的控制信号外;进而,也可通过图像识别的方式判别出车辆是否已经完全通过扫描,作为关闭射线的控制信号。判断车辆是否完全通过也可采用实施例1、2中的投影值判别方法,流程如图8所示,步骤如下:
1)车辆进入扫描区域,打开第一种辐射束开关,射线源发出第一种辐射束进行扫描,初始化空载标识变量Ctr0=0;
2)同实施例1算法一的步骤2)、3)、4)和5)
3)当Ctr0=0时,转入步骤4);当Ctr0=1时,转入步骤5);
4)当投影值Pi大于设定的阈值Thrd_0P时(Pi>Thrd_0P),则将Ctr0设为1,即Ctr0=1,返回步骤2);否则直接返回步骤2);其中阈值Thrd_0P为空载图像(不含物体的扫描图像)的投影值的平均值,若空载图像统计涨落较小时,Thrd_0P可设为0,若统计涨落较大可适当调高阈值。
5)当投影值Pi小于或等于设定的阈值Thrd_0P时(Pi<Thrd_0P),则说明图像又恢复成空载图像,车辆已经通过扫描,此时可关闭射线;否则,返回步骤2)。
以上,结合具体实施例对本发明的技术方案进行了详细介绍,所描 述的具体实施例用于帮助理解本发明的思想。本领域技术人员在本发明具体实施例的基础上做出的推导和变型也属于本发明保护范围之内。

Claims (15)

  1. 一种辐射源控制方法,其特征在于,包括:
    在辐射源开始发射第一辐射束扫描被检测物之后,通过辐射束探测器获取探测器信号Ii,其中i为探测器信号的顺序号,i=0,1,2,…,n;
    通过图像识别算法对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置;
    当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射第二辐射束;其中,
    所述分界位置是指被检测物的第一部分与第二部分之间的空隙,所述当前扫描位置与第s个探测器信号Is相对应。
  2. 如权利要求1所述的辐射源控制方法,其特征在于,其中所述对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置,包括:
    对探测器信号Ii的每个像素进行校正处理,得到校正图像ICi
    对校正图像ICi进行被检测物和背景的图像分割处理,得到分割图像IBi
    对分割图像IBi进行分界位置特征检测。
  3. 如权利要求2所述的辐射源控制方法,其特征在于,其中所述图像分割处理为二值化处理,所述二值化处理采用的阈值为μ*ICAir,其中,ICAir为校正图像ICi中空载部分的灰度值,μ为比例系数,0<μ<1;并且,将小于μ*ICAir的像素设为0,将大于等于μ*ICAir的像素设为1。
  4. 如权利要求2或3所述的辐射源控制方法,其特征在于,其中所述分界位置特征检测包括:
    ①如果i=0,初始化计数器Ctr0=0;
    ②对分割图像IBi进行垂直投影计算,得到投影值Pi,投影值Pi即为分割图像IBi中被检测物的像素的数量;
    ③用差分法计算投影值Pi的随i的变化梯度Gi
    ④如果Ctr=0,且投影值梯度Gi大于预定阈值Thrd_Gf,则判断为被检测物的第一部分的前沿到达扫描位置,将Ctr设为1,跳转至步骤②;如果 Ctr=0,而投影值梯度Gi小于或等于预定阈值Thrd_Gf,则直接跳转至步骤②;
    ⑤如果Ctr=1,且投影值梯度Gi小于预定阈值Thrd_Gb,则判断为被检测物的第一部分的后沿到达扫描位置,将Ctr设为2,跳转至步骤②;如果Ctr=1,而投影值梯度Gi大于或等于预定阈值Thrd_Gb,则直接跳转至步骤②;
    ⑥如果Ctr=2,且投影值梯度Gi大于上一梯度值Gi-1与预定阈值Thrd_IP之和,则判断为被检测物的分界位置到达扫描位置,流程结束;如果Ctr=2,而投影值梯度Gi小于或等于上一梯度值Gi-1与预定阈值Thrd_IP之和,跳转至步骤②。
  5. 如权利要求2或3所述的辐射源控制方法,其特征在于,其中所述分界位置特征检测包括:
    ①如果i=0,初始化投影最大值Pmax=0;
    ②对分割图像IBi进行垂直投影计算,得到投影值Pi,投影值Pi即为分割图像IBi中被检测物的像素的数量;
    ③计算当前投影值前m列投影值的均值Mi=(Pi+Pi-1+…+Pi-m+1)/m,其中m≥1,当m=1时,Mi=Pi;并且,当Mi>Pmax时,更新Pmax值,即令Pmax=Mi
    ④计算当前投影值前c列投影值的均值Ci=(Pi+Pi-1+…+Pi-c+1)/c,其中c≥1,当c=1时,Ci=Pi
    ⑤如果Ci<Pmax*Thrd_P,Thrd_P为判别阈值系数,则判断当前探测器信号Ii位于被检测物的第一部分与第二部分之间的分界位置,流程结束;否则,跳转至步骤②。
  6. 如权利要求1所述的辐射源控制方法,其特征在于,进一步地,所述图像识别算法还包括对探测器信号进行图像识别处理,以检测是否已经完成对被检测物的扫描;当检测到已完成被检测物扫描时,控制辐射源停止发射辐射束;其中,所述对探测器信号进行图像识别处理,以检测是否已经完成对被检测物的扫描,包括:
    对探测器信号Ii的每个像素进行校正处理,得到校正图像ICi
    对校正图像ICi进行被检测物和背景的图像分割处理,得到分割图 像IBi
    对分割图像IBi进行扫描位置对应空气且被检测物位于扫描位置下游的特征检测。
  7. 如权利要求6所述的辐射源控制方法,其特征在于,其中,
    所述图像分割处理为二值化处理,所述二值化处理采用的阈值为μ*ICAir,其中,ICAir为校正图像ICi中的空载部分的灰度值,μ为比例系数,0<μ<1;并且,将小于μ*ICAir的像素设为0,将大于等于μ*ICAir的像素设为1;
    其中所述扫描位置对应空气且被检测物位于扫描位置下游的特征检测包括:
    ①如果i=0,初始化计数器Ctr0=0;
    ②对分割图像IBi进行垂直投影计算,得到投影值Pi,投影值Pi即为分割图像IBi中被检测物的像素的数量;
    ③如果Crt0=0且投影值Pi大于预定阈值Thrd_0P,则判断扫描到被检测物,将Ctr0设为1,然后返回步骤②;
    ④如果Ctr0=0且投影值Pi小于或等于预定阈值Thrd_0P,则判断没有扫描到被检测物,返回步骤②;
    ⑤如果Ctr0=1且投影值Pi小于或等于预定阈值Thrd_0P,则判断当前扫描位置对应空气且被检测物位于扫描位置下游,流程结束;其中,
    所述预定阈值Thrd_0P为空载图像的投影值的平均值。
  8. 如权利要求1-7中任一项所述的辐射源控制方法,其特征在于,其中所述被检测物为车辆,所述第一部分为车辆的载客区域,所述第二部分为车辆的载货区域。
  9. 如权利要求1所述的辐射源控制方法,其特征在于,进一步地,所述图像识别算法还包括建立车辆模板库Tn,并将已经完成的车辆扫描图像Isumi与车辆模板进行匹配,其中n为模板个数;其中,所述车辆模板库Tn中的每个模板至少包含所述载客区域和所述分界位置。
  10. 如权利要求9所述的辐射源控制方法,其特征在于,其中所述扫描图像Isumi通过以下步骤获得:
    对探测器信号Ii的每个像素进行校正处理,得到校正信号ICi
    将已经获取的前i列校正信号IC1~ICi按先后顺序进行组合,并根据车辆移动速度进行几何校正,得到当前时刻已经完成的车辆扫描图像Isumi
  11. 如权利要求9或10所述的辐射源控制方法,其特征在于,其中所述匹配包括:
    从模板库Tn中选择有效模板VTm,m为有效模板的个数,有效模板VTm中的每个模板的图像宽度与扫描图像IMGi的宽度之差小于等于预设差值;
    将扫描图像Isumi与有效模板VTm中的每个模板逐一进行匹配,如果当前扫描图像Isumi与有效模板VTm中的第k个模板VTk的非相似度小于设定的阈值Thrd_S,则认为当前扫描图像Isumi与该第k个模板VTk匹配,则判断当前扫描位置位于车辆的分界位置。
  12. 一种速通式安检系统,其特征在于,包括:
    至少一个辐射源,用于发射至少两种辐射束;
    辐射束探测器,用于获取多个探测器信号Ii,其中i为探测器信号的顺序号,i=0,1,2,…,n;
    图像识别算法模块,用于对探测器信号进行图像识别处理,以检测当前扫描位置是否为被检测物的分界位置;
    辐射源控制模块,用于当检测到当前扫描位置为被检测物的分界位置时,控制辐射源开始发射指定辐射束;其中,
    所述分界位置是指被检测物的第一部分与第二部分之间的空隙,所述当前扫描位置与第s个探测器信号Is相对应。
  13. 如权利要求12所述的速通式安检系统,其特征在于,进一步地,所述图像识别算法模块还用于对探测器信号进行图像识别处理,以检测是否已经完成对被检测物的扫描;当检测到已完成被检测物扫描时,所述辐射源控制模块控制辐射源停止发射辐射束。
  14. 如权利要求12所述的速通式安检系统,其特征在于,其中所述被检测物为车辆,所述第一部分为车辆的载客区域,所述第二部分为车辆的载货区域。
  15. 如权利要求14所述的速通式安检系统,其特征在于,进一步地, 所述图像识别算法模块还用于建立车辆模板库Tn,并将已经完成的车辆扫描图像Isumi与车辆模板进行匹配,其中n为模板个数;其中,所述车辆模板库Tn中的每个模板至少包含所述载客区域和所述分界位置。
PCT/CN2016/083619 2016-02-24 2016-05-27 一种辐射源控制方法和速通式安检系统 WO2017143679A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610102115.0A CN105652332B (zh) 2016-02-24 2016-02-24 一种辐射源控制方法和速通式安检系统
CN201610102115.0 2016-02-24

Publications (1)

Publication Number Publication Date
WO2017143679A1 true WO2017143679A1 (zh) 2017-08-31

Family

ID=56489824

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/083619 WO2017143679A1 (zh) 2016-02-24 2016-05-27 一种辐射源控制方法和速通式安检系统

Country Status (3)

Country Link
CN (2) CN105652332B (zh)
SA (1) SA518392204B1 (zh)
WO (1) WO2017143679A1 (zh)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107664774A (zh) * 2017-09-19 2018-02-06 北京君和信达科技有限公司 辐射检查系统和方法
CN108614302A (zh) * 2018-07-11 2018-10-02 同方威视技术股份有限公司 辐射检查系统
CN109407162A (zh) * 2018-12-24 2019-03-01 同方威视技术股份有限公司 检查系统及成像方法
CN113205293A (zh) * 2021-04-23 2021-08-03 佛山沧科智能科技有限公司 一种汽车自动装货测量调度管控方法
CN113791459A (zh) * 2020-05-29 2021-12-14 同方威视技术股份有限公司 安检设备调试方法、装置、安检方法和设备
CN113835130A (zh) * 2020-06-23 2021-12-24 同方威视技术股份有限公司 自动行走式的检查装置和自动分车方法
CN113835131A (zh) * 2020-06-23 2021-12-24 同方威视技术股份有限公司 自动行走式的检查装置和自动分车方法
CN113947760A (zh) * 2021-12-20 2022-01-18 北京东方国信科技股份有限公司 一种绿色通道车辆的检测方法及装置

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106249307B (zh) * 2016-07-22 2018-12-28 同方威视技术股份有限公司 安全检查方法和系统
CN107228869A (zh) * 2017-06-29 2017-10-03 北京君和信达科技有限公司 辐射检查系统和辐射检查方法
CN107655906A (zh) * 2017-09-26 2018-02-02 同方威视技术股份有限公司 车辆安检系统、方法和控制器
CN108535138B (zh) * 2018-05-04 2024-04-16 鹤壁昊海电气有限公司 用于绿色通道车辆的比重检测装置及方法
CN109828310B (zh) * 2018-12-28 2024-05-03 同方威视技术股份有限公司 安检设备和安检方法

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050169421A1 (en) * 2004-01-30 2005-08-04 Muenchau Ernest E. Method and system for automatically scanning and imaging the contents of a moving target
CN103984035A (zh) * 2014-05-15 2014-08-13 北京君和信达科技有限公司 一种双模速通式移动目标辐射检查系统及方法
CN104090308A (zh) * 2014-07-22 2014-10-08 北京君和信达科技有限公司 对移动目标进行辐射扫描的系统、方法以及数据信息标签
CN204009098U (zh) * 2014-05-15 2014-12-10 北京君和信达科技有限公司 一种双模速通式移动目标辐射检查系统
CN204009100U (zh) * 2014-07-22 2014-12-10 北京君和信达科技有限公司 对移动目标进行辐射扫描的系统
CN104374785A (zh) * 2014-11-14 2015-02-25 北京君和信达科技有限公司 一种连续通过式辐射扫描系统和方法
CN104391339A (zh) * 2014-12-17 2015-03-04 同方威视技术股份有限公司 车型识别方法以及利用该方法的车辆快速检查系统
CN204314236U (zh) * 2014-11-14 2015-05-06 北京君和信达科技有限公司 一种连续通过式辐射扫描系统

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101019042B (zh) * 2004-03-01 2010-05-12 瓦润医药系统公司 通过双能量辐射扫描和缓发中子探测来检查物体
US7526064B2 (en) * 2006-05-05 2009-04-28 Rapiscan Security Products, Inc. Multiple pass cargo inspection system
CN100485710C (zh) * 2006-12-22 2009-05-06 四川川大智胜软件股份有限公司 利用数字图像处理技术识别车辆类型的方法
GB0803642D0 (en) * 2008-02-28 2008-04-02 Rapiscan Security Products Inc Drive-through scanning systems
CN102129732B (zh) * 2010-12-10 2013-07-17 山东申普交通科技有限公司 基于射频和图像识别组合监控的车辆监控方法
JP2013080326A (ja) * 2011-10-03 2013-05-02 Sony Corp 画像処理装置、画像処理方法及びプログラム
CN103150904A (zh) * 2013-02-05 2013-06-12 中山大学 一种基于图像特征的卡口车辆图像识别方法
US9086496B2 (en) * 2013-11-15 2015-07-21 Varian Medical Systems, Inc. Feedback modulated radiation scanning systems and methods for reduced radiological footprint
CN104361343B (zh) * 2014-10-30 2018-08-31 中盟科技有限公司 车辆类型识别方法及其装置

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050169421A1 (en) * 2004-01-30 2005-08-04 Muenchau Ernest E. Method and system for automatically scanning and imaging the contents of a moving target
CN103984035A (zh) * 2014-05-15 2014-08-13 北京君和信达科技有限公司 一种双模速通式移动目标辐射检查系统及方法
CN204009098U (zh) * 2014-05-15 2014-12-10 北京君和信达科技有限公司 一种双模速通式移动目标辐射检查系统
CN104090308A (zh) * 2014-07-22 2014-10-08 北京君和信达科技有限公司 对移动目标进行辐射扫描的系统、方法以及数据信息标签
CN204009100U (zh) * 2014-07-22 2014-12-10 北京君和信达科技有限公司 对移动目标进行辐射扫描的系统
CN104374785A (zh) * 2014-11-14 2015-02-25 北京君和信达科技有限公司 一种连续通过式辐射扫描系统和方法
CN204314236U (zh) * 2014-11-14 2015-05-06 北京君和信达科技有限公司 一种连续通过式辐射扫描系统
CN104391339A (zh) * 2014-12-17 2015-03-04 同方威视技术股份有限公司 车型识别方法以及利用该方法的车辆快速检查系统

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107664774A (zh) * 2017-09-19 2018-02-06 北京君和信达科技有限公司 辐射检查系统和方法
CN108614302A (zh) * 2018-07-11 2018-10-02 同方威视技术股份有限公司 辐射检查系统
CN109407162A (zh) * 2018-12-24 2019-03-01 同方威视技术股份有限公司 检查系统及成像方法
CN109407162B (zh) * 2018-12-24 2024-04-02 同方威视技术股份有限公司 检查系统及成像方法
CN113791459A (zh) * 2020-05-29 2021-12-14 同方威视技术股份有限公司 安检设备调试方法、装置、安检方法和设备
CN113835130A (zh) * 2020-06-23 2021-12-24 同方威视技术股份有限公司 自动行走式的检查装置和自动分车方法
CN113835131A (zh) * 2020-06-23 2021-12-24 同方威视技术股份有限公司 自动行走式的检查装置和自动分车方法
CN113205293A (zh) * 2021-04-23 2021-08-03 佛山沧科智能科技有限公司 一种汽车自动装货测量调度管控方法
CN113947760A (zh) * 2021-12-20 2022-01-18 北京东方国信科技股份有限公司 一种绿色通道车辆的检测方法及装置

Also Published As

Publication number Publication date
CN108897055A (zh) 2018-11-27
CN105652332B (zh) 2019-02-12
SA518392204B1 (ar) 2022-04-17
CN108897055B (zh) 2020-02-21
CN105652332A (zh) 2016-06-08

Similar Documents

Publication Publication Date Title
WO2017143679A1 (zh) 一种辐射源控制方法和速通式安检系统
US6873912B2 (en) Vehicle tracking system
US10527525B2 (en) Method and system for fast inspecting vehicle based on measure lengths
CN104391339B (zh) 车型识别方法以及利用该方法的车辆快速检查系统
US9934442B2 (en) Passenger counting device, passenger counting method, and program recording medium
RU2396512C2 (ru) Способ и система проверки движущегося объекта путем радиационного формирования изображения
US8068134B2 (en) Apparatus and method for predicting collision
CN101030256B (zh) 车辆图像分割方法和装置
KR100459476B1 (ko) 차량의 대기 길이 측정 장치 및 방법
CN117197141B (zh) 一种汽车零部件表面缺陷检测方法
CN105022095B (zh) 一种速通式移动目标辐射检查方法和系统
US20210396684A1 (en) Assembly for detecting defects on a motor vehicle bodywork
US8254632B2 (en) Detection of motor vehicle lights with a camera
US10395104B2 (en) Vehicle imaging system, vehicle imaging method and device, program, and recording medium
KR101898051B1 (ko) 다차선 차량 속도 측정 시스템
US20150198735A1 (en) Method of Processing 3D Sensor Data to Provide Terrain Segmentation
KR102062579B1 (ko) 영상 보정을 통해 그림자 및 빛 반사로 훼손된 차량번호판을 인식하는 차량번호판 인식 시스템
CN115240170A (zh) 一种基于事件相机的道路行人检测跟踪方法及系统
CN116228656A (zh) 车辆检查方法、系统及存储介质
EP4198553A1 (en) Method, device and program for estimating object reflectivity in a lidar system
CN108885262A (zh) 用于捕捉至少一个对象的方法、传感器设备的装置、传感器设备和具有至少一个传感器设备的驾驶员辅助系统
CN108426902B (zh) 一种基于视频的运动车辆位置检测方法
JP2013104783A (ja) 異物検知装置
KR20170075101A (ko) 강판 검사 장치 및 방법
US20240329216A1 (en) External environment recognition apparatus and method for adjusting parameters of recognition algorithm

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

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

Ref document number: 16891133

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 11/12/2018)

122 Ep: pct application non-entry in european phase

Ref document number: 16891133

Country of ref document: EP

Kind code of ref document: A1