SE1550006A1

SE1550006A1 - Method and system for categorization of a scene

Info

Publication number: SE1550006A1
Application number: SE1550006A
Authority: SE
Inventors: Amritpal Singh; Ulf Erlandsson
Original assignee: Viscando Ab
Priority date: 2015-01-07
Filing date: 2015-01-07
Publication date: 2016-06-14
Also published as: US10127458B2; CA2970911A1; EP3243165B1; SE538405C2; EP3243165A4; EP3243165A1; WO2016111640A1; CA2970911C; US20170344835A1

Description

scene, preferably also allowing for the possibility of extracting further information relating to the detected objects while still keeping the privacy at an in comparison high level.

SUMMARY OF THE INVENTION According to an aspect of the invention, the above is at least partly met by a computer-implemented method for classifying a moving object within a scene, the moving object being one of a plurality of different types of predetermined objects, the method comprising receiving an image stream comprising a plurality of images, each of the plurality of images being a captured representation of the scene and comprising depth information as to content within the scene, wherein the content of the scene comprises at least one physical foreground object and a physical background, processing the plurality of images to produce an equal plurality of depth maps of the scene, forrning height maps for each of the plurality of images based on a predeterrnined model of the background and the corresponding depth map, extracting the at least one physical foreground object from each of the plurality of images based on the corresponding height map, determining, for each of the plurality of images, a relative position of the at least one physical foreground object within the scene based on the height maps, deterrnining a first probability for the type of object within each of the images by matching the extracted physical foreground object with a predeterrnined set of different object types, Wherein each of the predetermined set of different object types are defined to have at least a length, a height and a width being within a predetermined range, deterrnining a second probability for the type of the object by determining a difference in the relative position for the object in at least two images of the plurality of images, converting the difference in relative position to a relative speed for the object, and matching the relative speed with speed profiles for the predeterrnined set of different object types, and de?ning the type of the foreground object based on the first and the second probability.

The general concept of the present invention is based on the fact that it may be possible to improve the classification of objects within a scene by determination of different information about the extracted object, and combine these pieces of information (e.g. a ?rst and a second piece of information about the object as indicated above) based on a probability level for each of the pieces of information. This will advantageously reduce any short term "noise" aspects that may result in error e. g. one of the pieces of information. However, when seen over time and when taking into account the reliability for each of the pieces of information a robust object Classification may be achieved. 10 15 20 25 30 This is speci?cally use?al when performing object Classification in a "mixed object" scene, e. g. where both a pedestrian and a vehicle are present. This will of course be further advantageous in scenes comprises a plurality of pedestrians and vehicles of different forms and shapes, e. g. including cars, trucks and bikes/bicycles.

It may also be possible to further improve the deterrnination of the type of object by including at least a third probability for the type of the obj ect, wherein the third probability is deterinined by matching the extracted foreground object with a set of prede?ned images for the plurality of different types of objects, and deterrnining a probability level, being a representation for the third probability, based on a resulting matching level.

It is advantageous, and within the scope of the invention, to store, for each of the plurality of images, data representing the type of the extracted object(s) and its relative position. Once the data is stored, e.g. within a database, it may for example be possible to track the objects within subsequent images of the image stream. It may also be possible to determine movement pattem for the different types of objects within the scene.

In the most general implementation of the invention, image data is received and processed. However, it may be possible to further control at least two cameras for acquiring the image stream, wherein each of the at least to cameras are arranged to capturing a predeterrnined spatially separated corresponding image of the scene. Each image pair will then need to be processed for deterrnining the depth information based on the corresponding images captured by the at least two cameras, for example using a known stereo/ 3D processing algorithm.

In some implementations of the invention it may be advantageous to determine the absolute location of the object(s) within the scene. This may for example be achieve by means of an absolute location deterrnined within the scene, where the relative locations for the objects based may be transforrned to absolute positions based on the absolute position within the scene. In a possible embodiment, the absolute position within the scene may be acquired from a GPS receiver arranged within the scene.

According to another aspect of the invention there is provided an image processing system for classifying a moving object within a scene, the moving object being one of a plurality of different types of predetermined objects, the image processing system comprising a control unit con?gured to receiving an image stream comprising a plurality of images, each of the plurality of images being a captured representation of the scene and comprising depth information as to content within the scene, wherein the content of the scene comprises at least one physical foreground object and a physical background, processing the 10 15 20 25 30 plurality of images to produce an equal plurality of depth maps of the scene, forming height maps for each of the plurality of images based on a predeterrnined model of the background and the corresponding depth map, extracting the at least one physical foreground object from each of the plurality of images based on the corresponding height map, deterrnining, for each of the plurality of images, a relative position of the at least one physical foreground object within the scene based on the height maps, deterrnining a first probability for the type of object within each of the images by matching the extracted physical foreground object with a predetermined set of different object types, wherein each of the predeterrnined set of different object types are defined to have at least a length, a height and a width being within a predeterrnined range, deterrnining a second probability for the type of the object by deterrnining a difference in the relative position for the object in at least two images of the plurality of images, converting the difference in relative position to a relative speed for the object, and matching the relative speed With speed profiles for the predeterrnined set of different object types, and defining the type of the foreground object based on the first and the second probability. This aspect of the invention provides similar advantages as discussed above in relation to the previous aspects of the invention.

In a preferred embodiment, the system further comprises at least two cameras wherein each of the at least to cameras are arranged to capturing a predeterrnined spatially separated corresponding image of the scene. In such an embodiment, the control unit thus further configured to control at least two cameras for acquiring the image stream, and to determine the depth information based on the corresponding images captured by the at least two cameras.

Within the context of the invention it may alternatively be possible to include a time-of-?ight (ToF) camera for acquiring the depth information as to the objects within the scene. In a possible implementation of the invention, a combination of "normal" 2D cameras are used together with at least one ToF camera According to a still further aspect of the invention there is provided a computer program product comprising a computer readable medium having stored thereon computer program means for operating an image processing system comprising a control unit, the image processing system configured for classifying a moving object within a scene, the moving object being one of a plurality of different types of predetermined objects, the computer program product comprising code for receiving an image stream comprising a plurality of images, each of the plurality of images being a captured representation of the scene and comprising depth information as to content within the scene, wherein the content 10 15 20 25 30 of the scene comprises at least one physical foreground object and a physical background, code for processing the plurality of images to produce an equal plurality of depth maps of the scene, code for forrning height maps for each of the plurality of images based on a predeterrnined model of the background and the corresponding depth map, code for extracting the at least one physical foreground object from each of the plurality of images based on the corresponding height map, code for deterrnining, for each of the plurality of images, a relative position of the at least one physical foreground object within the scene based on the height maps, code for deterrnining a first probability for the type of object within each of the images by matching the extracted physical foreground object with a predeterrnined set of different object types, wherein each of the predeterrnined set of different object types are defined to have at least a length, a height and a width being within a predeterrnined range, code for determining a second probability for the type of the object by deterrnining a difference in the relative position for the object in at least two images of the plurality of images, converting the difference in relative position to a relative speed for the object, and matching the relative speed With speed profiles for the predeterrnined set of different object types, and code for defining the type of the foreground object based on the first and the second probability. Also this aspect of the invention provides similar advantages as discussed above in relation to the previous aspects of the invention.

The control unit preferably including a micro processor or any other type of computing device. Similarly, a so?ware executed by the control unit for operating the inventive wireless communication device may be stored on a computer readable medium, being any type of memory device, including one of a removable nonvolatile random access memory, a hard disk drive, a ?oppy disk, a CD-ROM, a DVD-ROM, a USB memory, an SD memory card, or a similar computer readable medium known in the art. Accordingly, operation of the wireless communication device may be at least partly automated, implemented as e.g. software, hardware and a combination thereof.

Further features of, and advantages with, the present invention will become apparent when studying the appended claims and the following description. The skilled addressee realize that different features of the present invention may be combined to create embodiments other than those described in the following, without departing from the scope of the present invention. 10 15 20 25 30 BRIEF DESCRIPTION OF THE DRAWINGS The various aspects of the invention, including its particular features and advantages, will be readily understood from the following detailed description and the accompanying drawings, in which: Fig. 1 illustrates an image processing system according to the invention arranged for collecting and processing image data of a scene; Fig. 2 provides a conceptual illustration of the image processing system illustrated in Fig. 1; Fig. 3 is a ?ow chart illustrating the exemplary steps for operating the system according to the invention, and Fig. 4 depicts details of object classification according to the invention.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS The present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which currently preferred embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided for thoroughness and completeness, and fully convey the scope of the invention to the skilled addressee. Like reference characters refer to like elements throughout.

Referring now to the drawings and Fig. 1 in particular, there is depicted an image processing system 100 arranged in an elevated position as to a ground/street level, in the illustrated embodiment af?xed to an elongated light post 102. The elevated position of the image processing system 100 provides for the possibility of monitoring a scene comprising a plurality of different physical objects. Specifically, in the disclosed embodiment the scene comprises a plurality of different types of physical foreground objects including pedestrians 104, different types of vehicles, such as a truck 106 and a car 108, as well as a plurality of bicycles 110. The scene also comprises a background, for example including a street 112, trees 114, etc. It should be noted that the image processing system 100 may be affixed to or integrated into any form of elevated structure. As such, it could be possible to arrange the image processing system 100 onto e. g. a bridge, an overpass, etc. The image processing system 100 could possibly be integrated with a sign post or any other form of arrangement. The image processing system 100 could be stationary of mobile, connected to the main grid or battery powered. 10 15 20 25 30 The truck 106 and the car 108 travel along the street 112, in opposite directions in different lanes, Whereas the pedestrians 104 and the bicycles 110 shares the pavement space at the outsides of the street traveling at mixed directions at each of the pavements.

In the illustrated embodiment, the image processing system 100 collects image data for a preselected portion of the scene. The size of the preselected portion of the scene may be dependent on the capacity of the image processing system 100, the resolution of the cameras, etc. The collection of image data and possible image processing performed by the image processing system 100 may as exemplified in Fig. 1 be communicated, e. g. Wirelessly or Wired, to a remote server (not shown), for example over the Intemet. Any type of Wired or wireless communication protocol could be possible in accordance to the invention.

Tuming now to Fig. 2, Which conceptually illustrates the image processing system 100 as shown in Fig. 1. The image processing system 100 comprises a first (left) 202 and a second (right) camera 204 having optical lenses connected thereto, respectively. The image processing system 100 comprises a casing 206, preferably configured for handling outdoor use, the casing holding the cameras 202, 204 as Well as processing and communication ?anctionality for the image processing system 100. Typically, the image processing system 100 comprises a control unit 208 arranged in communication With the cameras 202, 204 as Well as a communication module 210 (for allowing communication With e.g. the above mentioned remote server) and a memory element 212 (e. g. for interrnediate storage of image data). It should be understood that the cameras 202, 204 may not necessarily be provided as separate elements, but may be integrated as one single unit provided With a frst and a second set of optical lenses arranged to a first and a second image sensor (i.e. provided for the cameras 202, 204 respectively).

As understood from the above, the image processing system 100 is configured for stereo collection of image data using the first and the second camera 202, 204. To improve the possibility of creating useful three dimensional image data it is desirable to separate the first and the second camera 202, 204 With a predeterrnined distance, the distance being dependent on the desired implementation. Furthermore, within the scope of the invention, it should be understood that it could be possible to use other forms of cameras for image collection, such as using a time-o f-flight (ToF) camera. Also, the general concept of the invention may be implemented based on image data captured using "external" cameras (i.e. not necessarily including the cameras themselves). That is, the invention may for example be implemented as a server solution Where the server receives image data from e.g. 10 15 20 25 30 remotely arranged Cameras. As such, at least some of the processing performed could be implemented in a distributed manner, e. g. partly as a "cloud" solution.

During operation of the image processing system 100, With further reference to Fig. 3, the process starts by the control unit 208 receiving, S1, an image stream from the first and the second camera 202, 204. The control unit 208 subsequently process, S2, the image stream for forrning a three dimensional representation of the scene monitored by the cameras 202, 204. The three dimensional representation is typically formed based on depth information extracted based on the predeterrnined separation of the ?rst 202 and the second 204 camera, thereby forming depth maps for each image of the image stream (i.e. pair of images from the cameras 202, 204).

Based on the depth map and a predetermined model of the background it is possible to forrn, S3, height maps for each of the image pair. The background model could for example be an "empty" scene where no foreground objects are present. The background model could also be deterrnined in other manners as Would be readily understood by the skilled addressee.

The height maps are subsequently used for extracting a physical foreground object from the scene, such as a pedestrian 104, a car 106 or a bicycle 110. The extracted foreground objects are typically defined to have a "boundary box" closely surrounding the object. The boundary box may typically be rectangular but may also be defined in a more freeforrn manner.

For each of the image pair a relative position for the foreground object is deterrnined, S5. The relative position is defined as being relative to e.g. one or a plurality of predetermined "points" Within the scene. However, it may of course be possible to make the position "absolute" by correlating one or a plurality of points Within the scene to e.g. be a real GPS position, or similar.

For each extracted object Within the scene a first probability as to the type of object is determined, S6. In a preferred embodiment of the invention a set of object types are prede?ned. Each of the set of the predefined object types are de?ned to have at least a length, a height and a width being within a predeterrnined range. The foreground object being extracted from the scene is accordingly matched to the set of predefined object types, and a probability is deterinined for the extracted foreground object as compared to the predefined object types. As understood, the set of prede?ned object types typically includes a set of height, width and length parameters defining the pedestrian, the bicycle, the car and the truck.

The size (e.g. the boundary box as mentioned above) of the extracted foreground object is 10 15 20 25 30 thus compared to the predefmed parameters for each of the type of objects for deterrnining the first probability.

A second probability is deterrnined, S7, where the extracted foreground object essentially is "tracked" between subsequent image pairs of the image stream. In accordance to the invention a difference in the relative position is determined for the extracted foreground object, i.e. from one image pair to the next. Accordingly, it may be possible to, for each of the predefined object types, form a set of "speed" parameters/profile defining how fast the speci?c object type likely will move between subsequent image pairs. The tracking will generally also include determination of the direction of movement of the extracted object.

Finally, the type of the foreground object is deterrnined, S8, by combining the first and the second probability, possibly with further information for increasing the probability of correctly determining the object type for the extracted foreground object.

This is further exempli?ed in Fig. 4 showing details of the object classification method according to the invention. Specifically, the ?owchart illustrates the above discussed geometric matching classification, e.g. including matching of size and "moving object positions" for subsequent image pairs with predefmed object type parameters (i.e. 404 and 406). In addition, a further, e. g. possibly defined as a third probability may be deterrnined by performing an image matching 408 between the extracted foreground object and representative image data stored in a database 410. The image matching may preferably take into account the direction of movement of the extracted object, for example for use in improving the selection of the "correct" image to match with, i.e. as the stored images to match with may be annotated to include information as to an aspect angle for the stored image. The third probability for the object type may typically be combined with the first and the second probability.

It should be understood that the combination of the first, second and third probability may be weighted, e.g. more weight could for example be given to the height, width and length parameter matching as compared to the remaining two probabilities. Any type of weighting may be possible and is within the scope of the invention. Also, it should be understood that the combined probability may be accumulated over time, i.e. by allowing further geometric matching classifications and/or image based classifications to be made for the same object. That is, it will likely be preferred to continue to track the object as long as the object is visible within the scene. Thereby, the likelihood of determining the correct object type will be increased. 10 15 20 25 30 10 Once the object type has been deterrnined, i.e. being one of the prede?ned types of object, this knowledge may be used for a numerous different applications. For example, the image processing system 100 could be used in a traf?c counting scenario comprising tracking mixed object types for categorizing the behavior for each of the object types. Similarly, the image processing system 100 according to the invention could be used for controlling traffic patterns, i.e. to control traf?c lights in a city, for example by adapting time given for passing a street. For example, at peak hours with a high density of bicycles, the bicycles could be given more passing time as compared to cars, possibly improving the overall traffic environment within the city.

In addition, the control functionality of the present disclosure may be implemented using existing computer processors, or by a special purpose computer processor for an appropriate system, incorporated for this or another purpose, or by a hardwired system.

Embodiments within the scope of the present disclosure include program products comprising machine-readable media for carrying or having machine-executable instructions or data structures stored thereon. Such machine-readable media can be any available media that can be accessed by a general purpose or special purpose computer or other machine with a processor. By way of example, such machine-readable media can comprise RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to carry or store desired program code in the form of machine-executable instructions or data structures and which can be accessed by a general purpose or special purpose computer or other machine with a processor. When information is transferred or provided over a network or another communications connection (either hardwired, Wireless, or a combination of hardwired or wireless) to a machine, the machine properly views the connection as a machine-readable medium. Thus, any such connection is properly terrned a machine-readable medium.

Combinations of the above are also included within the scope of machine-readable media.

Machine-executable instructions include, for example, instructions and data which cause a general purpose computer, special purpose computer, or special purpose processing machines to perform a certain function or group of functions.

Although the figures may show a sequence the order of the steps may differ from what is depicted. Also two or more steps may be performed concurrently or with partial concurrence. Such variation will depend on the software and hardware systems chosen and on designer choice. All such Variations are within the scope of the disclosure. Likewise, software implementations could be accomplished with standard programming techniques 10 ll with rule based logic and other logic to accomplish the various connection steps, processing steps, comparison steps and decision steps. Additionally, even though the invention has been described With reference to specific exemplifying embodiments thereof, many different alterations, modiñcations and the like Will become apparent for those skilled in the art.

Further, a single unit may perforrn the functions of several means recited in the claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting to the claim. Furthermore, in the claims, the Word "comprising" does not exclude other elements or steps, and the inde?nite article "a" or "an" does not exclude a plurality.

Variations to the disclo sed embodiments can be understood and effected by the skilled addressee in practicing the claimed inventíon, from a study of the drawings, the disclosure, and the appended claims. The person skilled in the art realizes that the present invention is not limited to the preferred embodiments.

Claims

1 PATENTKRAV 1. En datorimplementerad metod for att klassificera en typ av rorligt objekt (104, 106, 108, 110) i en seen, dar det rOrliga objektet är en av ett flertal olika typer avforutbestamda objekt, varvid metoden innefattar att: 1. motta (Si) en bildstrom innefattande ett flertal bilder, dar var och en av 5 namnda flertal bilder är en inhamtad representation av scenen och innefattar djupinformation avseende innehallet i scenen, varvid innehallet i scenen innefattar atminstone ett fysiskt forgrundsobjekt och en fysisk bakgrund; 2. bearbeta (S2) namnda flertalet bilder for att skapa en lika stor mangd av djupkartor for scenen; - skapa (S3) hajdkartor for vart och ett av namnda flertal bilder baserat pa en forutbestamd modell av bakgrunden och den motsvarande djupkartan; 3. extrahera (S4) atminstone ett fysiskt forgrundsobjekt fran var och en av namnda flertalet av bilder baserat pa den motsvarande hojdkartan; 4. bestamma (S5), for var och en av namnda flertal bilder, en relativ position hos niimnda atminstone ett fysiskt fOrgrundsobjekt i scenen baserat pa hojdkartorna; 5. bestamma (S6) en forsta sannolikhetsnivâ for typen av objekt inom var och en av bilderna genom att matcha det extraherade fysiska forgrundsobjektet med en forutbestamd uppsattning av olika objekttyper, varvid var och en av den forutbestamda uppsattningen av olika objekttyper är definierade att ha atminstone en langd, en hojd och en bredd som är inom ett forutbestamt omfang; 6. bestamma (S7) en andra sannolikhetsnivà fOr typen av objekt genom att bestamma en skillnad i den relativa positionen for objektet i atminstone tva bilder av namnda flertal bilder, omvandla skillnaden i relativ position till en relativ hastighet for objektet och matcha den relativa hastigheten med hastighetsprofiler for den forutbestamda uppsattningen av olika objekttyper; och 7. definiera (S8) typen av forgrundsobjektet baserat pa en kombination av den forsta och den andra sannolikhetsnivan. 2. Metod enligt krav 1, varvid de olika typerna av forutbestamda objekt innefattar en fotgangare (104), en cykel (110) och ett fordon (106, 108). 2 3. Metod enligt nagot av kraven 1 och 2, varvid typen av forgrundsobjekt vidare definiera baserat pa en tredje sannolikhetsniva fOr den typ av objektet, varvid den tredje sannolikheten bestams genom att: 1. matcha det extraherade forgrundsobjektet med en uppsattning fordefinierade bilder for flera olika typer av objekt, och 2. bestamma en sannolikhetsnivk som är en representation for den tredje sannolikheten, baserat pa en resulterande matchande niva. 4. Metod enligt nagot av kraven 1 och 2, varvid bestamningen av den forsta och den andra sannolikhetsnivan definieras medelst en geometrisk klassificerare. 5. Metod enligt krav 3, varvid bestamningen av den tredje sannolikhetsnivan definieras =deist en bildbaserad klassificerare. 6. Metod enligt nagot av kraven 1 och 2, vidare innefattande att: 1. bestamma en tredimensionell gransbox som omger det extraherade fcirgrundsobjektet. 7. Metod enligt nagot av kraven 1 och 2, vidare innefattande att: - folja det extraherade forgrundsobjektet genom efterfoljande bilder i bildstrOmmen. 8. Metod enligt nagot av foregaende krav, vidare innefattande att: 1. lagra, for var och en av namnda flertal bilder, data i en databas som representerar typen av det extraherade objektet och dess relativa position. 9. Metod enligt nagot av foregaende krav, vidare innefattande att: 1. styra atminstone tva kameror (202, 204) for att inhamta bildstrommen, varvid var och en av namnda atminstone tva kameror är anordnade att infanga flera forbestamt 30 rumsligt separerade bilder av scenen, och 2. bestamma djupinformationen baserat pa bildema som inhamtats med de atminstone tva kameroma. 3 10. Metod enligt nagot av foregaende krav, vidare innefattande att: 1. bestamning en absolut position for det extraherade forgrundsobjektet baserat pa en absolut position for kminstone en position i scenen. 11. Ett bildbehandlingssystem (100) som anpassat for att klassificera en typ av rOrligt objekt (104, 106, 108, 110) i en seen, dar det rOrliga objektet är en av ett flertal olika typer av forutbestamda objekt, varvid bildbehandlingssystemet innefattande en styrenhet (208) som konfigurerad att: 1. motta en bildstrom innefattande ett flertal bilder, dar var och en av namnda flertal bilder är en inhamtad representation av scenen och innefattar djupinformation avseende innehallet i scenen, varvid innehallet i scenen innefattar atminstone ett fysiskt forgrundsobjekt och en fysisk bakgrund; 2. bearbeta namnda flertalet bilder for att skapa en lika stor mangd av djupkartor for scenen; - skapa hojdkartor for vart och ett av namnda flertal bilder baserat pa en forutbestamd modell av bakgrunden och den motsvarande djupkartan; 3. extrahera atminstone ett fysiskt forgrundsobjekt fran var och en av namnda flertalet av bilder baserat pa den motsvarande hojdkartan; 4. bestamma, for var och en av namnda flertal bilder, en relativ position hos 20 namnda atminstone ett fysiskt forgrundsobjekt i scenen baserat pa hojdkartorna; 5. bestamma en forsta sannolikhetsnivâ for typen av objekt inom var och en avbilderna genom att matcha det extraherade fysiska forgrundsobjektet med en forutbestamd uppsattning av olika objekttyper, varvid var och en av den fOrutbestamda uppsattningen av olika objekttyper är definierade att ha atminstone en langd, en hojd och en bredd som är inom ett forutbestamt omfang; 6. bestamma en andra sannolikhetsniva for typen av objekt genom att bestamma en skillnad i den relativa positionen for objektet i kminstone tva bilder av namnda flertal bilder, omvandla skillnaden i relativ position till en relativ hastighet for objektet och matcha den relativa hastigheten med hastighetsprofiler for den forutbestamda uppsattningen av olika objekttyper; och 7. definiera typen av forgrundsobjektet baserat pa en kombination av den forsta och den andra sannolikhetsnivan. 4 12. System (100) enligt krav 11, vidare innefattande: 1. atminstone tva kameror (202, 204), varvid var och en av namnda atminstone tva kameror anordnats att infanga flera forbestamt rumsligt separerade bilder av scenen, varvid styrenheten vidare konfigurerats for att: - styra namnda atminstone tva kameror for att inhamta bildstrommen, och 2. bestamma djupinformationen baserat pa bilderna som inhamtats med de atminstone tva kamerorna. 13. System (100) enligt krav 12, vidare innefattande: - ett positionsgranssnitt for att motta en absolut position for atminstone en position i scenen, varvid styrenheten vidare konfigurerats for att: 1. bestamma en absolut position for det extraherade forgrundsobjektet baserat pa den absoluta positionen. 14. System (100) enligt krav 13, vidare innefattande en GPS-mottagare som är ansluten till positionsgranssnittet och som tillhandahaller den absoluta positionen fcir systemet. 15. En datorprogramprodukt innefattande ett datorlasbart medium som har lagrat darpa datorprogramorgan for att handha ett bildbehandlingssystem (100) innefattande en styrenhet (208), varvid bildbehandlingssystemet konfigurerats for klassificering av en typ av rorligt objekt (104, 106, 108, 110) i en seen, varvid det rorliga objektet är ett av ett flertal olika typer av forutbestamda objekt, varvid datorprogramprodukten innefattar: - kod for att motta en bildstrom innefattande ett flertal bilder, dar var och en av namnda flertal bilder är en inhamtad representation av scenen och innefattar djupinformation avseende innehallet i scenen, varvid innehallet i scenen innefattar atminstone ett fysiskt forgrundsobjekt och en fysisk bakgrund;

1. kod for att bearbeta namnda flertalet bilder for att skapa en lika stor mangd 30 av djupkartor for scenen;

2. kod for att skapa hojdkartor for vart och ett av namnda flertal bilder baserat pa en forutbestamd modell av bakgrunden och den motsvarande djupkartan;

3. kod for att extrahera atminstone ett fysiskt forgrundsobjekt fran var och en av namnda flertalet av bilder baserat pa den motsvarande hojdkartan;

4. kod for att bestamma, for var och en av namnda flertal bilder, en relativ position hos namnda atminstone ett fysiskt forgrundsobjekt i scenen baserat pa hOjdkartorna;

5. kod for att bestamma en forsta sannolikhetsniva for typen av objekt inom var och en av bildema genom att matcha det extraherade fysiska forgrundsobjektet med en forutbestamd uppsattning av olika objekttyper, varvid var och en av den forutbestamda uppsattningen av olika objekttyper är definierade att ha kminstone en langd, en hojd och en bredd som är inom ett forutbestamt omfang;

6. kod for att bestamma en andra sannolikhetsniva for typen av objekt genom att bestamma en skillnad i den relativa positionen for objektet i atminstone tva bilder av namnda flertal bilder, omvandla skillnaden i relativ position till en relativ hastighet for objektet och matcha den relativa hastigheten med hastighetsprofiler for den forutbestamda uppsattningen av olika objekttyper; och

7. kod for att definiera typcn av forgrundsobjektet bascrat pa en kombination av den forsta och den andra sannolikhetsnivan. zot 214 102 100 /