EP0396432A2

EP0396432A2 - Monitoring apparatus

Info

Publication number: EP0396432A2
Application number: EP90304912A
Authority: EP
Inventors: Michael Dalgleish; David Fine; Bob Moreton; Andrew Watkins; John Jones
Original assignee: GOLDEN RIVER Ltd
Current assignee: GOLDEN RIVER Ltd
Priority date: 1989-05-05
Filing date: 1990-05-04
Publication date: 1990-11-07
Also published as: GB2231753A; CA2016088A1; EP0396432A3; GB8910419D0

Abstract

Monitoring apparatus for monitoring traffic flow in a tunnel is disclosed comprising a plurality of cameras 10, 12, 14 each connected to a camera controller 60, 62, 64. Each camera controller includes an image memory for periodically storing still images from the camera associated therewith. A plurality of incident detectors 40 - 45; 46 - 51 are associated with each camera 10, 12, and lie within the field of view of that camera. The incident detectors detect the passage of vehicles in a carriageway of the tunnel and determine whether or not a traffic incident eg congestion or an accident has occurred in accordance with an algorithm. When an incident has occurred an incident detection signal is generated which is communicated to the camera controllers via a central controller 66. The image memory of the relevant camera controller is responsive to the detection signal to identify and freeze a plurality of images prior to the incident which are then passed to the central controller, so that the circumstances leading up to the incident can be reviewed.

Description

This invention relates to monitoring apparatus, particularly, but not exclusively for monitoring traffic flow in a tunnel.
Monitoring apparatus for intruder detection is known, for example that manufactured by the Vision Research Company Limited under the trade name Pixstore 256. Such apparatus generally comprises a video camera connected to a monitor screen via a control unit which is further connected to an intruder detection device eg an infra red, micro wave, ultrasonic or perimeter type device, the control unit actuating an alarm and freezing the image from the video camera at the instant of detection thus allowing a 'snap shot' picture of the intrusion.
It is a disadvantage of such a system that it is only capable of responding to and recording an incident at the time the incident occurs and is thus unsuitable for providing information as to how the incident arose. The system is thus unsuitable for monitoring traffic related incidents.
According to the invention in a first aspect, there is provided monitoring apparatus comprising a camera, an image memory for storing images from the camera, an incident detector for generating an incident detection signal when an incident occurs, the image memory being responsive to the detection signal whereby a plurality of said images prior to the incident are identified; and means for reviewing said prior images.
Preferably, the monitoring apparatus is used for detecting a traffic incident, preferably in a tunnel, the camera being a video camera providing images to a camera controller which stores the images periodically in an image memory. The incident detector preferably comprises a road sensor, for sensing when a vehicle passes in the field of view of the camera and a processor for determining the occupancy of the road and, from this information, if an incident has occurred, the incident detector then generating the incident detection signal.
Other preferred features of the invention are mentioned in dependent Claims 2 to 20.
It is a further disadvantage of the prior art apparatus that the control unit is disposed at a central monitoring location, servicing a plurality of video cameras so that malfunction of the unit causes the whole monitoring system to be disabled.
According to the invention in a second aspect there is provided monitoring apparatus comprising a plurality of monitoring stations each having a camera and a local camera controller, the camera controller having an image memory for storing images from the camera; and a remote central controller connected to the camera controllers for selectively receiving said stored images.
According to the invention in a third aspect, there is provided a method of determining the existence of a traffic incident comprising the steps of measuring the average speed of vehicles passing a sensor in a first time period, measuring the average speed of vehicles passing the sensor in second time period shorter than the first time period, calculating the difference between the average speeds for the two time periods and generating an incident detection signal when the difference is above a predetermined threshold.
An embodiment of the invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic diagram showing the disposition of cameras and incident detectors of an embodiment of the invention in one carriage way of a road tunnel.
Figure 2 is a schematic diagram showing an embodiment of the invention.
Figure 3 is a schematic diagram of a camera controller of the embodiment of Figure 2.
Figure 4 is a schematic diagram of the data conversion unit of the embodiment of Figure 2.
Figure 5 is a schematic diagram of a site controller of the embodiment of Figure 2.
Figure 6 is a schematic diagram of a vehicle detection module of the embodiment of Figure 2.

Referring to the figures, an embodiment of monitoring apparatus according to the invention is shown.
The monitoring apparatus is arranged for use in traffic monitoring, to monitor and detect a traffic incident on a highway and, specifically, in a road tunnel.
As shown in Figure 1, a plurality of video cameras 10, 12, 14 are attached at spaced intervals to the wall of a tunnel, each camera having a field of view 22, 24, 26 which overlaps with the field of view of the preceding camera. The cameras monitor one carriage way 30 having two lanes 32, 34 in the tunnel 20 and each camera 10, 12, 14 has associated therewith a plurality of vehicle sensors 40 - 53 .... of which sensors 40 - 45 are associated with camera 10 and sensors 46 - 51 are associated with camera 12. As described below, the sensing of an incident by one or more of the sensors associated with the camera will cause the retention of images showing both the incident and the circumstances leading up to the incident. The sensors in each lane are separated by a distance D which is chosen according to conditions and is preferably 60m.
A block diagram of an embodiment of monitoring apparatus of the invention as a whole is shown in Figure 2.
Each camera 10, 12, is connected to a local camera controller 60, 62, 64 each controller processing images received from its respective camera. All camera controllers are connected to a central controller 66 remote from the camera sites via a data bus 70, the central controller having a supervising computer 72, with associated hard disc storage and hard copy facilities, a monitor 74 and a data conversion unit 76 shown in more detail in Figure 4 and formed of standard electronic circuits providing D/A and A/D conversion, modulation and demodulation and filtering of video signals and instructions. The unit 76 acts as an I/O and processing interface between the computer 72, monitor 74 and camera controllers 60 - 64.
Camera controller 60 is shown in more detail in Figure 3 and includes an A/D converter 80 for converting the analogue video image signal from the video camera 10 to a digital equivalent and providing an output digital signal to a central processing unit 82. The CPU 82 stores frames of the digital video image signal at intervals in a memory stack 84 typically having capacity for 32 frames with a resolution 192 x 300 pixels x 64 grey levels, implying a memory requirement of approximately 24 K bytes/frame. The interval period may be either constant or variable and is preferably of at least one second duration, so that the memory stack provides a record of past events of at least 32 seconds duration.
The CPU 82 is further connected to a data interface for transmitting/receiving signals to/from central controller 66 via bus 70. Each camera controller has a unique address and upon an instruction signal from the central computer 72 tagged with the address, can transmit real time video images direct from the video camera 10 or can transmit all or part of frame store 84. The camera controllers only transmits on instruction from central computer 66. The CPU 82 is also responsive to an incident detection signal, from central control 66 on bus 70 or from an incident detector (described below) on data line 88 to freeze the contents of frame memory 84 in the event of an incident being detected, so that the contents of the frame store, which records past images prior to the detected incident are retained for transmission to the central controller 66, thus providing a record of the circumstances leading up to the detected incident. Optionally, the CPU can move a frame pointer so that some of the frames labelled F are stored and future, post incident images are stored higher up the stack, thus allowing retention of past events and at the same time continued monitoring of current events.
The frame memory 84 may optionally be formed as a robust detatchable cartridge eg of the type disclosed in co-pending European patent application No 89303333.2, this allowing removal and retrieval of image data held in the cartridge in the event of system failure. The cartridge may be formed from physical and thermal shock resistant materials, for example a polycarbonate case having epoxy resin potting, so that retrieval of the cartridge when the camera controller has suffered all or partial destruction will still be possible. Use of a detatchable cartridge also allows for corroboration, after an incident, between the content of the frame memory and the image data transmitted to the central controller 66.
The central controller 66 is further connected, via a data bus 90, to a plurality of incident detectors 92, 94, 96. Each incident detector comprises a site controller 100, 102 104. Each site controller has connected thereto three vehicle detection modules (VDMs) 120, 122, 124; 126; 128; 130;...... each of which contains signal processing circuitry for a pair of the vehicle sensors 40, 41 ...., the pair being connected to the respective VDM. Each site controller is further connected to an associated one of the camera controllers via signal lines 140, 142, 144.
The vehicle detection module 120 is shown in more detail in Figure 6. The road sensor 40 comprises a square four turn inductive loop 200 which is connected to the VDM 120 via an isolating transformer 210. The VDM includes an alternating current source 220 having a frequency of approximately 60 kHz and a frequency measuring circuit comprising a programmable down counter 230, an elapse counter 240 connected to counter 230 by line 235 and driven by high speed clock (10 MHz) 250 for measuring the duration of the oscillation count and a CPU 260. Each sensor 40, 41 has its own alternating current source 220, the frequency measuring circuit being connected to one or other of the sensors by means of switch 270 under control of CPU 260 in a time division manner. Preferably, the CPU 260 switches between sensors 40, 41 at one millisecond intervals.
In use, the sensor 40 is placed in the road lane either attached on the road surface or sunk into the road surface. The inductance of the loop 200 will drop when a metal bodied vehicle passes over. This in turn will affect the oscillation frequency of the circuit comprising the loop 200 and source 220. The frequency measuring circuit measures this frequency by counting down a selected number of oscillations of the alternating voltage signal; when the counting operation is being performed, a signal on line 235 changes level thus providing a start/stop signal to counter 240 which measures the time duration of the count, this giving the period of the alternating signal and thus its frequency. The count value is passed to CPU 260 via bus 255.
Preferably, the CPU controls the counter to output a start/stop signal on line 235 after a plurality of oscillations (eg 8, 16 or 32) selected by user configurable switches 237, to improve resolution accuracy.
The CPU 260 processes the counter information and compares the derived frequency measurement with a threshold, producing a true/false signal indicating presence adjacent the loop 40 of a vehicle. The signal is sent to the site controller 100 on bus 265.
The site controller 120 is shown in more detail in Figure 5 and is of stand alone construction, based on the MARKSMAN 600 traffic management controller manufactured by the applicants. The site controller 120 has a I/O circuit 300 which received frequency information signals on buses 265 - 267 from respective VDMs 120, 122, 124. The I/O circuit also provides an output on bus 140 to camera controller 60.
The I/O circuit and all other functions of the site controller are controlled by a CPU 310 to which is further connected a keyboard/display 315, to allow on site initialisation and input/output of data from the site controller, a data cartridge 320 and interface 325 preferably of a type disclosed in European patent application No 89303333.2, for storing incident and occupancy data, a ROM/RAM 355 for storing CPU programmes and operational data and a network interface 330 which provides a communications link with data bus 90.
The network interface 330 includes a processor 335 and two port universal asyncronous receiver transmitter (UART) 340 for data transfer. A switched bypass bus 350 having a plurality of electromechanical relays is further provided, the relays being biased closed but held open by the site controller CPU 310. The bus 350 acts to 'short circuit' the network interface 330 in the event of power failure of the site controller 120 or when the site controller cannot make sense of signals being transmitted through UART 340, so that failure of one site controller will not affect the operation of others connected to data bus 90 downstream of the failed site controller.
The CPU has random access and read only memories 355 for internal data storage and for storing control and incident identification programmes. The site controller 120 is connected to a power supply and also has a local backup supply in the form of a rechargable battery (not shown) for use in the event of a general power failure.
In use, the site controller receives the vehicle presence information from the VDMs 120, 122, 124 and from this calculates the degree of 'occupancy' and 'density' of any one sensor by vehicles. Occupancy is defined as the number of consecutive seconds that a vehicle has been sensed as present by the sensor.
Density is defined as the percentage time in a given time interval that the loop has sensed the presence of a vehicle or vehicles. This information is then used to calculate if a traffic incident has occurred in accordance with an algorithm. Such algorithms are known to those skilled in the art, as exemplified by the high occupancy (HIOCC) algorithm developed by the Transport and Road Research Laboratory (TRRL) as disclosed in TRRL supplementary reports Nos 775 (Automatic incident detection, experience with two TRRL algorithm HIOCC; J F Collins 1983) and 526 (Automatic incident detection - TRRL algorithms HIOCC and PATREG; J F Collins, C M Hopkins and J A Martin 1979).
The HIOCC algorithm as disolosed in the above documents, the contents of which are incorporated herein by reference, operates by detecting stationary or slow moving vehicles to indicate a traffic queue caused by an incident or by congestion. It looks for several consecutive seconds of high detector occupancy to detect queues and incidents in high traffic flows. A programme in accordance with the flow diagram of Figure 3 of report 526 is stored in ROM in site controller 120 and CPU 310 processes the occupancy data from VDMs 120 - 124 in accordance with the programmed algorithm. The resultant occupancy, density and incident data is stored locally in data cartridge 320 and is also sent to the central controller 66 via data bus 90.
When the algorithm detects a traffic incident, an incident detection signal is sent both to the central controller and to the camera controller 60 associated with the site controller 100. The incident detection signal causes the camera controller to freeze a predetermined number of images in the frame store 84 as previously described, thus providing a stored record of the circumstances leading up to the incident as detected by the incident detector.
In addition to the HIOCC algorithm, the site controller uses a speed (as opposed to occupancy) based algorithm using two adjacent loop sensors eg 41, 43 in any one lane. Such an algorithm provides additional information concerning slow moving vehicles - which are, in themselves, a traffic hazard. Furthermore, by basing analysis on speed, speeding violations may also be detected.
An example of a suitable algorithm is as follows:
Two measurement intervals T1 and T2 (between one minute and twenty four hours depending on occupancy) are chosen, interval T1 representing a relatively longer period than T2. Using adjacent loop detectors 41, 43 the average speed in each interval T1, T2 is calculated and updated as each vehicle passes, giving average speeds S1, S2. The difference between these speeds (S1-S2) gives an indication of short term speed variation away from the long term average and if over a predetermined threshold SD gives an indication of an incident. Furthermore, if S1 or S2 or the instantaneous vehicle speed fall outside predetermined high or low speed thresholds THH, THL, this also gives an indication of a probable incident. Examples of suitable parameters are T1 = 1 hour, T2 = 6 minutes, SD = 5 kph, THH = 200 kph, THL = 20 kph.
The site controllers 100, 102, 104 form nodes of a local area network (LAN) having a standard format and operating protocols, each node passing messages along the data bus 90. Each controller has a unique network identifier, with identifiers being reserved for 'all stations' called and the central control 66.
The topology of the network is a daisy chain with out going messages being passed to the end of the line and incoming messages being passed back to the central controller 66. Each site controller is responsible for passing messages along the line when a character arrives it is buffered until the whole message is complete. It is then retransmitted.
All site controllers examine messages, discard corrupt messages, accept those with matching addresses and pass on others.
When a site controller has a message of its own for the central controller 66, it tests the status of the incoming line from the previous site controller. If the line is busy, the site controller will continue to re-tranmsit data from the previous site controller until the line is clean, at which point the site controller will commence transmission of its own message. During this time any incoming messages from the outlying site controllers will be buffered in the site controller UART for re-transmission at the earliest opportunity.
Site controllers are assigned unique addresses and all messages from site controllers to the central controller 66 are tagged with this address. Messages from the central controller to site controllers are either 'all stations' to all site controllers or 'addressed' to individual site controllers.
Messages are transmitted along the LAN as the data field of a network packet, packets having the following format:

1. Packet header
2. Destination address
3. Source address
4. Control flag
- A: Acknowledgement
- B: Text message
- C: incident detection/alarm message
- D: Status message
5. Field check sum
6. Data box size
7. Data block

The packet receiving protocol for each site controller is as follows:

1. Each complete message is re-transmitted along the line.
2. Incoming characters are also placed into the incoming message buffer.
3. When a full packet is received the unit compares the destination address with its own node id.
4. If the id and address do not match no further action is taken.
5. If the id and address match then the packet is interpreted, involving the following:
- A. The checks are miscalculated
- B. If the check sum is incorrect a bad packet error is flagged
- C. The response to a bad packet error is to do nothing and allow the source to time out and resend
- D. If the checks are incorrect, the data block is passed to the CPU for action and an acknowledgement (ACK) packet is sent to the central controller.

The packet sending protocol of the site controllers is as follows:

1. The messages are formatted by the site controller CPU into a packet having the form noted above.
2. The packet is placed in an outgoing packet queue and the count of re-tries set to zero.
3. The packet is transmitted.
4. If an ACK packet is received from the target node/central controller with the correct packet ID then the packet has been successfully transmitted and is removed from the queue.

If no reply is received before a predetermined time out then the packet is re-sent and the count of retries for this packet is incremented.
If the count of time out re-tries reaches the user specified maximum then a message time out error is flagged.
During normal operation when no incident detections are present the central controller 66 will poll the status of each site controller by sending a request for a status packet. In reply to the status request the units will respond with a data packet giving the following parameters:

1. Occupancy (the number of consecutive seconds for which a sensor has been found to be occupied).
2. Density: the percentage of time at a given interval for which the loop has been occupied.
3. Incident detection status.
4. Loop status.
5. Miscellaneous.

When an incident detection signal is generated, the sensing unit sends an incident detection packet to the central controller 66. The central controller 66 acknowledges receipt of the packet (otherwise the packet is re-sent). The central controller then acts to freeze the memory store for the relevant camera/camera controller. In parallel, an incident detection signal is sent direct from the site controller to its associated camera controller.
While the embodiment of the invention as described above has been applied to a road traffic sensing system, this is not to be construed as limitative. For example, the invention may be used in a driver's cab of a train, the incident detector being responsive to an automatic warning system (AWS) 'line occupied' signal showing, for example, when a train has gone through a danger signal. Alternatively, the incident detector could be a sensor connected to the front buffers of the train so that data is stored on impact with another vehicle or object on the track. In such circumstances, the data cartridge could be made to 'black box' standards.

Claims

1. Monitoring apparatus comprising:
a camera (10) and an incident detection (40) for generating an incident detection signal when an incident occurs, characterised by additionally comprising an image memory (60) for storing images from the camera, by the image memory being responsive to the detection signal whereby a plurality of images prior to the incident are identified, and means (66) for reviewing said prior images.

2. Apparatus as claimed in Claim 1, wherein the image memory is divided into a plurality of fields each field storing an image frame.

3. Apparatus as claimed in Claim 2 wherein the time period between frames is variable.

4. Apparatus as claimed in Claim 2 or 3 wherein the image memory comprises a first in first out stack.

5. Apparatus as claimed in any one of the preceding claims further comprising at least one further camera, each further camera having a respective image memory for storing images from the camera, a plurality of further incident detections at least one incident detector being operably associated with each camera.

6. Apparatus as claimed in Claim 5 where further comprising a central control means for receiving incident detection signals from the detectors and prior images from the cameras.

7. Apparatus as claimed in Claim 1 wherein the incident detector comprises at least one vehicle sensor for generating a vehicle sensing signal and processing means for receiving said sensing signal and for determining from the signal if an incident has occured.

8. Monitoring apparatus comprising a plurality of monitoring stations each having a camera and a localised camera controller, the camera controllers each having an image memory for storing images from the camera; and a remote central controller connected to the camera controllers for selectively receiving said stored images.

9. A method of determining the existence of a traffic incident comprising the steps of measuring the average speed of vehicles passing a sensor in a first time period, measuring the average speed of vehicles passing the sensor in a second time period shorter than the first time period, calculating the difference between the average speeds for the two time periods and of generating an incident detection signal when the difference is above a predetermined threshold.

10. A method as claimed in Claim 9 further comprising the step of comparing the average speeds to first and second thresholds and generating an incident detection signal when the average speeds fall above the first threshold or below the second threshold.