CN107368837A - Method for checking object and object test equipment - Google Patents
Method for checking object and object test equipment Download PDFInfo
- Publication number
- CN107368837A CN107368837A CN201610320029.7A CN201610320029A CN107368837A CN 107368837 A CN107368837 A CN 107368837A CN 201610320029 A CN201610320029 A CN 201610320029A CN 107368837 A CN107368837 A CN 107368837A
- Authority
- CN
- China
- Prior art keywords
- confidence level
- parallax
- apart
- projection
- subject area
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Measurement Of Optical Distance (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of method for checking object for being used to detect the object in identification image or video and the object test equipment using the method for checking object.The method for checking object includes:Obtain the anaglyph and range image for including object to be detected;Based on the anaglyph, obtain the parallax object with the first confidence level, and be based on the range image, obtain have the second confidence level apart from object;And based on the parallax object with the first confidence level and it is described with the second confidence level apart from object, it is determined that the subject area detected, wherein, first confidence level and second confidence level indicate respectively the parallax object and the confidence level for belonging to same object apart from object.
Description
Technical field
The present invention relates to image processing field, more particularly it relates to which a kind of be used to detect identification figure
The method for checking object and the object detection dress using the method for checking object of picture or the object in video
Put.
Background technology
In recent years, in computer vision field, detection and knowledge for the object in image or video
Other technology is significant and its application is also more and more extensive.For example, in intelligent driving accessory system
In, the accurately detection for the object in Driving Scene is most important.
Currently, it has been proposed that the object detection technique based on anaglyph, it is such as three-dimensional by clustering
Parallax pixel in the anaglyph that camera obtains carrys out detection object.However, it is possible to due to shade, dim light
According to etc. reason cause parallax point discontinuous, or even on some parts on some objects no parallax point or
Only few parallax point.Therefore, it is difficult to accurately completely detect the object in scene in these cases.
In addition, it is also proposed that the object detection technique based on range image, it by such as laser radar by being launched
Detection signal and the detection signal that is reflected back by destination object, the distance of the reflection object distance of acquisition
Image carrys out detection object.However, because range image has the noise pixel as caused by the impurity in air,
It is difficult to accurately realize the object detection in scene.
A kind of accordingly, it is desirable to provide method for checking object for being used to detect identification image or the object in video
And the object test equipment using the method for checking object, its anaglyph and distance map based on scene
Picture and both confidence levels for object detection, realize the accurate detection of object in scene.
The content of the invention
In view of the above problems, the present invention provides a kind of pair for being used to detect identification image or the object in video
Object test equipment as detection method and using the method for checking object.
According to one embodiment of present invention, there is provided a kind of method for checking object, including:Acquisition includes
The anaglyph and range image of object to be detected;Based on the anaglyph, acquisition has the first confidence
The parallax object of degree, and be based on the range image, obtain with the second confidence level apart from object;
And based on the parallax object with the first confidence level and the distance pair with the second confidence level
As, it is determined that the subject area detected, wherein, first confidence level and second confidence level difference
Indicate the parallax object and the confidence level for belonging to same object apart from object.
In addition, method for checking object according to an embodiment of the invention, wherein it is described based on it is described away from
From image, obtain includes with the second confidence level apart from object:Based on the range image, it is determined that tool
Have second confidence level apart from line segment;And based on described apart from line segment, obtaining has described second
Confidence level apart from object, wherein, it is described apart from line segment include by the straight line preset distance away from horizontal direction
In the range of the line segment that forms of first kind point and/or the line segment that is made up of the continuous second class point on oblique line, and
And the coordinate based on the first kind point and/or the second class point determines second confidence level.
In addition, method for checking object according to an embodiment of the invention, wherein described be based on the tool
Have the first confidence level parallax object and it is described have the second confidence level apart from object, it is determined that detecting
Subject area include:By on the plane of delineation of the parallax Object Projection to the range image, obtain
The parallax object of projection;Match the parallax object of the projection with it is described apart from object;Compare described first
Confidence level and second confidence level, determine that the height in first confidence level and second confidence level is put
Reliability;And based on the high confidence level, extend or split the parallax object of the projection and the distance
One in object.
In addition, method for checking object according to an embodiment of the invention, wherein described be based on the tool
Have the first confidence level parallax object and it is described have the second confidence level apart from object, it is determined that detecting
Subject area also include:Extend or split the parallax object of the projection with it is described in object
After one, with described apart from the case that object is misaligned, repetition is based on the parallax object of the projection
The extension or segmentation of the high confidence level, until the parallax object of the projection overlaps with described apart from object,
It is determined that detected corresponding to the parallax object of the projection of coincidence with described apart from the region of object to be described
Subject area.
In addition, method for checking object according to an embodiment of the invention, in addition to:Based on detecting
Subject area, update the characteristic information of object in the subject area.
According to another embodiment of the invention, there is provided a kind of object test equipment, including:Image obtains
Part is taken, for obtaining the anaglyph and range image that include object to be detected;Object acquisition part,
For based on the anaglyph, obtaining the parallax object with the first confidence level, and based on it is described away from
From image, obtain with the second confidence level apart from object;Subject area obtaining widget, for based on institute
State parallax object with the first confidence level and it is described with the second confidence level apart from object, it is determined that inspection
The subject area measured, wherein, first confidence level and second confidence level indicate respectively described regard
Poor object and the confidence level for belonging to same object apart from object.
In addition, object test equipment according to another embodiment of the invention, wherein the object acquisition
Part is based on the range image, it is determined that with second confidence level apart from line segment;And based on institute
State apart from line segment, obtain with second confidence level apart from object, wherein, it is described apart from line segment bag
Include the line segment that is made up of the first kind point in the straight line predetermined distance range away from horizontal direction and/or by oblique line
On the continuous second class point line segment that forms, and based on the first kind point and/or the second class point
Coordinate determines second confidence level.
In addition, object test equipment according to another embodiment of the invention, wherein the subject area
Obtaining widget includes:Parallax Object Projection unit, for by the parallax Object Projection to the distance map
On the plane of delineation of picture, the parallax object of projection is obtained;Object matching unit, for matching the projection
Parallax object with it is described apart from object;Confidence level comparing unit, for first confidence level and
Second confidence level, determine the high confidence level in first confidence level and second confidence level;With
And subject area correction unit, for based on the high confidence level, extending or splitting the parallax of the projection
Object and one in object.
In addition, object test equipment according to another embodiment of the invention, wherein extending or splitting
The parallax object of the projection with after one in object, the parallax object of the projection with
It is described apart from the case that object is misaligned, the subject area correction unit repeats to be based on the high confidence
The extension or segmentation of degree, until the parallax object of the projection overlaps with described apart from object, the object
Region acquisition portion part determine corresponding to overlap the projection parallax object with it is described apart from the region of object
For the subject area detected.
In addition, object test equipment according to another embodiment of the invention, in addition to:Characteristics of objects
Part is updated, for based on the subject area detected, the feature for updating object in the subject area to be believed
Breath.
The object detection side according to an embodiment of the invention for being used to detect identification image or the object in video
Method and the object test equipment using the method for checking object, its anaglyph and distance based on scene
Image and both confidence levels for object detection, realize the accurate detection of object in scene.
It is to be understood that foregoing general description and following detailed description are both exemplary, and
And it is intended to provide the further explanation of claimed technology.
Brief description of the drawings
The embodiment of the present invention is described in more detail in conjunction with the accompanying drawings, it is of the invention above-mentioned and its
Its purpose, feature and advantage will be apparent.Accompanying drawing is used for providing entering one to the embodiment of the present invention
Step understands, and a part for constitution instruction, is used to explain the present invention together with the embodiment of the present invention,
It is not construed as limiting the invention.In the accompanying drawings, identical reference number typically represent same parts or
Step.
Fig. 1 is the functional block diagram for illustrating object test equipment according to an embodiment of the invention;
Fig. 2 is the flow chart for illustrating method for checking object according to an embodiment of the invention;
Fig. 3 is further to illustrate in method for checking object according to an embodiment of the invention apart from object and regard
The flow chart of poor object acquisition processing;
Fig. 4 is further to illustrate in method for checking object according to an embodiment of the invention apart from object and regard
The schematic diagram of poor object acquisition processing;
Fig. 5 is that the subject area further illustrated in object test equipment according to an embodiment of the invention obtains
Take the block diagram of part;
Fig. 6 is further to illustrate detection object region in method for checking object according to an embodiment of the invention
Obtain the flow chart of processing;
Fig. 7 is further to illustrate detection object region in method for checking object according to an embodiment of the invention
Obtain the schematic diagram of processing;
Fig. 8 is the flow chart for further illustrating method for checking object according to an embodiment of the invention;
Fig. 9 is the functional block diagram for further illustrating object test equipment according to an embodiment of the invention;
Figure 10 is the exact functionality for further illustrating object test equipment according to an embodiment of the invention
Block diagram;And
Figure 11 is the schematic diagram for illustrating object detection systems according to an embodiment of the invention.
Embodiment
Become apparent in order that obtaining the object, technical solutions and advantages of the present invention, below with reference to accompanying drawings in detail
Thin description is according to example embodiment of the invention.Obviously, described embodiment is only the one of the present invention
Section Example, rather than the whole embodiments of the present invention, it should be appreciated that the present invention is not by described herein
The limitation of example embodiment.Based on the embodiment of the present invention described in the present invention, those skilled in the art exist
All other embodiment obtained by not paying in the case of creative work should all fall into the guarantor of the present invention
Within the scope of shield.
Hereinafter, embodiments of the invention will be described in detail with reference to the attached drawings.
First, the object test equipment according to the present invention is described into reference picture 1.Fig. 1 is diagram according to this
The functional block diagram of the object test equipment of the embodiment of invention.As shown in figure 1, the reality according to the present invention
Applying the object test equipment 10 of example includes image acquisition component 200, object acquisition part 300 and target area
Domain obtaining widget 400.Image acquisition component 200, object acquisition portion in the object test equipment 10
Part 300 and subject area obtaining widget 400 can be by the central processing list of the object test equipment 10
First (CPU) configuration.Alternately, described image obtaining widget 200, object acquisition part 300 and right
As region acquisition portion part 400 can also be (all by the specialized processing units in the object test equipment 10
Such as application specific integrated circuit (ASIC)).That is, described image obtaining widget 200, object acquisition
Part 300 and subject area obtaining widget 400 for example can by such as hardware, software, firmware and it
Any feasible combination configuration.It is easily understood that Fig. 1 only shows the object test equipment 10
In certainly can be with the closely related module of the present invention, object recognition equipment according to embodiments of the present invention
Including other modules, inputoutput unit, communication component etc..
Specifically, described image obtaining widget 200 be used for obtain comprising object to be detected anaglyph and
Range image.That is, different from only obtain the region comprising object to be identified anaglyph or
One of range image comes the situation of perform detection, described image obtaining widget according to an embodiment of the invention
200 obtain by least two image capture apparatus (such as unshowned anaglyph acquisition equipment and distance
Image capture apparatus) capture the anaglyph and range image that include object to be detected.
In one embodiment of the invention, the anaglyph from anaglyph acquisition equipment (for example,
Binocular camera).The range image comes from range image acquisition equipment (for example, range laser radar).
The anaglyph acquisition equipment and the range image acquisition equipment can obtain with described image thereafter
Part 200 is physically separated, or is physically located at same position even on same casing internal.
The anaglyph acquisition equipment and the range image acquisition equipment and described image obtaining widget thereafter
In the case that 200 are physically separated, the anaglyph acquisition equipment and the range image acquisition equipment
Further the anaglyph of acquisition and range image are sent to thereafter via wired or wireless mode
Part.In the anaglyph acquisition equipment and the range image acquisition equipment and described image thereafter
It is described to regard in the case that obtaining widget 200 is physically located at same position even on same casing internal
Difference image acquisition equipment and the range image acquisition equipment via internal bus will obtain anaglyph and
Range image is sent to part thereafter.In another embodiment of the present invention, according to the implementation of the present invention
The object test equipment of example can not also include the anaglyph acquisition equipment and the range image captures
Device, but received via wired or wireless mode from other electronic equipments to contain object to be identified
Region anaglyph and range image.
The object acquisition part 300 is used to be based on the anaglyph, obtains with the first confidence level
Parallax object, and be based on the range image, obtain with the second confidence level apart from object.At this
In one embodiment of invention, the object acquisition part 300 is based on the anaglyph, utilizes cluster
(cluster) processing obtains the parallax object with the first confidence level.In addition, the object acquisition part
300 are based on the range image, it is determined that with the second confidence level apart from line segment, it is described apart from line segment bag
Include the line segment that is made up of the first kind point in the straight line predetermined distance range away from horizontal direction and/or by oblique line
On continuous second class point form line segment.The object acquisition part 300 is based further on the distance
Line segment, obtained using segmentation (segment) processing with second confidence level apart from object.At this
In the embodiment of invention, first confidence level and second confidence level indicate respectively the parallax object
With the confidence level for belonging to same object apart from object.As will be described in detail, based on parallax
The distance and number of parallax point on object, determine first confidence level;Based on the first kind point and/or
The coordinate of the second class point determines second confidence level.
The subject area obtaining widget 400 be used for based on the parallax object with the first confidence level with
And it is described have the second confidence level apart from object, it is determined that the subject area detected.The one of the present invention
In individual embodiment, the subject area obtaining widget 400 is by the parallax Object Projection to the distance map
On the plane of delineation of picture, the parallax object of projection is obtained;Match the parallax object of the projection with it is described away from
From object;Compare first confidence level and second confidence level, determine first confidence level and institute
State the high confidence level in the second confidence level;And the high confidence level is based on, extends or splits the projection
Parallax object and one in object.Via the parallax pair for extending or splitting the projection
As with after one in object, judge the parallax object of the projection with it is described apart from object
Whether overlap.If the parallax object of the projection is with described misaligned apart from object, the target area
Domain obtaining widget 400 repeat the parallax object of extension based on the high confidence level or the segmentation projection with
The processing of one in object, until the projection parallax object with it is described apart from object weight
Close, and determine corresponding to overlap the projection parallax object with it is described apart from the region of object be institute
State the subject area detected.
Fig. 2 is the flow chart for illustrating method for checking object according to an embodiment of the invention.By above by reference to
The object test equipment according to an embodiment of the invention 10 of Fig. 1 descriptions performs the method for checking object.
Method for checking object according to an embodiment of the invention comprises the following steps:
In step s 201, the anaglyph and range image for including object to be detected are obtained.As described above,
Described image obtaining widget 200 according to an embodiment of the invention is obtained and caught by least two image capture apparatus
That obtains includes the anaglyph and range image of object to be detected.Hereafter, processing enters step S202.
In step S202, based on the anaglyph, the parallax object with the first confidence level is obtained,
And it is based on the range image, obtains with the second confidence level apart from object.As described above, according to
The object acquisition part 300 of embodiments of the invention is based on the anaglyph, and obtaining has first
The parallax object of confidence level, and the range image is based on, obtain the distance pair with the second confidence level
As.First confidence level and second confidence level indicate respectively the parallax object and described apart from right
Confidence level as belonging to same object.Hereinafter, the object is described in further detail with reference to the accompanying drawings to obtain
Take the flow handled apart from object and parallax object acquisition that part 300 performs.Hereafter, processing enters step
Rapid S203.
In step S203, based on the parallax object with the first confidence level and described with second
Confidence level apart from object, it is determined that the subject area detected.As described above, the implementation according to the present invention
The subject area obtaining widget 400 of example is based on the parallax object with the first confidence level and institute
The interaction apart from object with the second confidence level is stated, correction has the subject area of different confidence levels,
To obtain the subject area detected.Hereinafter, the subject area is described in further detail with reference to the accompanying drawings
The concrete configuration of obtaining widget 400, and the detection object region being executed by it obtain the flow of processing.
Therefore, in the object detection according to an embodiment of the invention dress above by reference to Fig. 1 and Fig. 2 general introductions
Put with method for checking object, based on the parallax object with the first confidence level and with the second confidence level
The interaction apart from object, correction with different confidence levels subject area, it is right in scene to realize
The accurate detection of elephant.
Hereinafter, method for checking object middle-range according to an embodiment of the invention is described into for reference picture 3 and Fig. 4
The flow chart and schematic diagram handled from object and parallax object acquisition.
Fig. 3 is further to illustrate in method for checking object according to an embodiment of the invention apart from object and regard
The flow chart of poor object acquisition processing.Fig. 4 is further to illustrate object inspection according to an embodiment of the invention
The schematic diagram handled in survey method apart from object and parallax object acquisition.As shown in figure 3, according to the present invention
Embodiment method for checking object in apart from object and parallax object acquisition processing comprise the following steps:
The anaglyph and distance map for including object to be detected are obtained in the step S201 that reference picture 2 describes
As after, processing enters the step S2021 shown in Fig. 3.
In step S2021, based on the anaglyph, the parallax object with the first confidence level is obtained.
Shown in reference picture 4, Fig. 4 (A) is shown captures dress by the anaglyph of such as binocular camera
Put the anaglyph 401 of capture.Described image obtaining widget 200 according to an embodiment of the invention is based on
The anaglyph 401, the parallax object 404 with the first confidence level is obtained using clustering processing.Specifically
Ground, the confidence level C of parallax object1Below equation (1) can be used to calculate:
Wherein, m represents the number of parallax point on parallax object, ziIt is the distance of i-th of parallax point, zavg
It is the average distance of all parallax points on parallax object.
Referring back to Fig. 3, after step S2021, processing enters step S2022.
In step S2022, based on the range image, it is determined that the distance with second confidence level
Line segment.
Shown in reference picture 4, Fig. 4 (B) is shown to be caught by the range image of such as range laser radar
Obtain the range image 402 of device capture.Described image obtaining widget 200 according to an embodiment of the invention
Based on the range image 402, it is determined that with second confidence level apart from line segment 403.Fig. 4's
(B) in, the line segment being made up of the first kind point in the straight line predetermined distance range away from horizontal direction is shown
With the line segment being made up of the continuous second class point on oblique line.Specifically, first kind point forms horizontal linear line
The confidence level C of section2Below equation (2) can be used to calculate:
Wherein, n represents the number of the point on straight-line segment, yiIt is i-th point of vertical coordinate, yavgDirectly
On line line segment somewhat vertical coordinate average value.
The confidence level C3 of the second horizontal diagonal segment lines of class point composition can be calculated as below:
For example, it is ax+by+c=0 to be fitted the diagonal segment lines with Hough transformation, then the point in range image
(x0,y0) to the distance d of the line segment be:
So the second class point forms the confidence level C of horizontal diagonal segment lines3Below equation (4) can be used to calculate:
Wherein, p represents the number of the point in diagonal segment lines, diIt is the distance of the i-th point of line segment to fitting,
davgIt is the average distance for the line segment for arriving fitting in diagonal segment lines a little.
Referring back to Fig. 3, after step S2022, processing enters step S2023.
In step S203, based on described apart from line segment, the distance pair with second confidence level is obtained
As.
Shown in reference picture 4, described image obtaining widget 200 according to an embodiment of the invention is based on described
Apart from line segment 403, using dividing processing obtain with the second confidence level apart from object 405.The distance
Object 405 has the confidence level of identical second with described apart from line segment 403.
More than, reference picture 3 and Fig. 4 describe method for checking object middle-range according to an embodiment of the invention
Handled from object and parallax object acquisition.Hereinafter, reference picture 5 to Fig. 7 is further described by according to this
Subject area obtaining widget in the object test equipment of the embodiment of invention, and obtained by the subject area
Part is taken to be based on obtaining the processing in detection object region apart from object and parallax object.
Fig. 5 is that the subject area further illustrated in object test equipment according to an embodiment of the invention obtains
Take the block diagram of part.As shown in figure 5, target area in object test equipment according to an embodiment of the invention
Domain obtaining widget 400 further comprises parallax Object Projection unit 4001, object matching unit 4002, put
Reliability comparing unit 4003 and subject area correction unit 4004.
The parallax Object Projection unit 4001 is used for the parallax Object Projection to the range image
The plane of delineation on, obtain the parallax object of projection.
The object matching unit 4002 be used to matching the parallax object of the projection with it is described apart from object.
The confidence level comparing unit 4003 is used for first confidence level and second confidence level,
Determine the high confidence level in first confidence level and second confidence level.
The subject area correction unit 4004 is used to be based on the high confidence level, extends or splits the throwing
The parallax object of shadow and one in object.
Described with further reference to Fig. 6 and Fig. 7 performed by the subject area obtaining widget 400 shown in Fig. 5
The acquisition processing of detection object region.Fig. 6 is further to illustrate object detection according to an embodiment of the invention
Detection object region obtains the flow chart of processing in method;Fig. 7 is reality of the further diagram according to the present invention
Apply the schematic diagram that detection object region in the method for checking object of example obtains processing.
As shown in fig. 6, the step of step S202 or reference picture 3 that are described in reference picture 1 are further described
Parallax object with the first confidence level is obtained in S2023 and with the second confidence level after object,
Processing enters step S2031.
In step S2031, by the plane of delineation of the parallax Object Projection to the range image, obtain
Take the parallax object of projection.As described above, the parallax Object Projection unit 4001 is by the parallax object
Project on the plane of delineation of the range image, obtain the parallax object of projection.
Shown in reference picture 7, Fig. 7 (A) shows that the parallax Object Projection unit 4001 will be such as figure
Parallax object 404 shown in 4 (C) is projected on the plane of delineation of the range image, the throwing of acquisition
The parallax object 701 of shadow.
Referring back to Fig. 6, after step S2031, processing enters step S2032.
In step S2032, match the parallax object of the projection with it is described apart from object.As described above,
The object matching unit 4002 match the parallax object of the projection with it is described apart from object.
Shown in reference picture 7, the object matching unit 4002 is for Fig. 7 (A) projection shown
Parallax object 701 and Fig. 7 (B) show apart from object 702, (that is, (D) of Fig. 4 is shown
Apart from object 405) perform matching.Determine that what Fig. 7 (C) showed includes horizontal range object and projection
Anaglyph matching region 703.
Referring back to Fig. 6, after step S2032, processing enters step S2033.
In step S2033, first confidence level and second confidence level, described is determined
High confidence level in one confidence level and second confidence level.As described above, the confidence level comparing unit
4003 are used for first confidence level and second confidence level, determine first confidence level and institute
State the high confidence level in the second confidence level.Hereafter, processing enters step S2033.
In step S2033, based on the high confidence level, the parallax object of the projection is extended or split
With one in object.As described above, the subject area correction unit 4004 is used to be based on
The high confidence level, extend or split the parallax object of the projection and one in object.
In reference picture 7 describes detection object region acquisition processing, it is assumed that second the putting apart from object
Reliability is higher than the first confidence level of the parallax object.As shown in fig. 7, due to described apart from the of object
Two confidence levels are higher than the first confidence level of the parallax object, and in the region 703 of matching, distance
Object 702 covers the projection 701 of the two sections of parallax objects shown in Fig. 7 (A), and is also connected with
The projection 701 of two sections of parallax objects, therefore subject area correction unit 4004 is based on the height
Confidence level it is described apart from object, described apart from the parallax object projected described in object extension.Namely
Say, increase parallax point near object described, obtain regarding for the extension as shown in Fig. 7 (D)
Poor pixel region 704.
It is easily understood that method for checking object not limited to this according to an embodiment of the invention, but can
It is higher than the situation of second confidence level apart from object with the first confidence level including the parallax object,
And the projection of the parallax object covers larger range of situation than the described of matching apart from object.
Specifically, it is assumed that parallax object is projected as L1, parallax object confidence level is C1;It is apart from object
L2, it is C apart from object confidence level2.The subject area correction unit 4004 is used to be based on the high confidence
Degree, extending or splitting the parallax object of the projection and one in object includes situations below.
1) L is worked as1<L2, and C1<C2, then increase parallax point apart from object based on described, regarded described in extension
Situation shown in (D) of poor object, i.e. reference picture 7.
2) L is worked as1<L2, and C1>C2, then the scope based on the parallax object, split described apart from right
As.
3) L is worked as1>L2, and C1<C2, then based on the scope apart from object, the parallax pair is split
As.
4) L is worked as1>L2, and C1>C2, then the scope based on the parallax object, extend described apart from right
As.
Referring back to Fig. 6, after step S2034, processing enters step S2035.
In step S2035, it is determined that extending or splitting the parallax object of the projection with described apart from right
After one as in, whether the parallax object of projection overlaps with described apart from object.
If obtaining negative decision in step S2035, i.e., parallax object and the distance of described projection
Object is still misaligned, then the processing returns to step S2033, to continue the parallax object based on the projection
With the high confidence level object in object, the parallax object of the projection is performed with described apart from right
The interactivity correction of elephant.
If on the contrary, obtain positive result in step S2035, i.e., the parallax object of described projection with
Described to be overlapped apart from object, then processing enters step S2036.
In step S2036, it is determined that corresponding to coincidence the projection parallax object with described apart from right
The region of elephant is the subject area detected.
Shown in reference picture 7, after the matching and correction that are performed by Fig. 7 (A) to (D), obtain
The subject area 706 that must be detected.
That compares that the interaction based on anaglyph and range image in Fig. 7 (E) finally obtains detects
Subject area 706 and be based only on the parallax object 404 (Fig. 4 (C) shown in) of anaglyph determination
Be based only on range image determination apart from object 405 (Fig. 4 (D) shown in), it is seen that according to this
The method for checking object of the embodiment of invention overcome based on anaglyph determine parallax object 404 when by
Cause the discontinuous defect of parallax point in reasons such as shade, low-lights, obtain complete detection object (figure
The truck shown in 7 (E)).In addition, method for checking object according to an embodiment of the invention is also gram
The noise pixel as caused by the impurity in air when having taken the parallax object 405 based on range image determination
Defect.
Fig. 8 is the flow chart for further illustrating method for checking object according to an embodiment of the invention.Fig. 8
Shown step S801 to S803 is identical with step S201 to the S203 that reference picture 2 describes, and herein will
Omit its repeated description.
Method for checking object according to an embodiment of the invention shown in Fig. 8 determines detection in step S803
After the subject area arrived, processing enters step S804.
In step S804, based on the subject area detected, the spy of object in the subject area is updated
Reference ceases.Specifically, after the object detection result of t frames is obtained, according to an embodiment of the invention
Method for checking object can be updated based on the object detection result before t frames in the subject area
The characteristic information of object.The characteristic information of the object such as size including object, position and it is based on
Translational speed of the object that the change of position determines etc..
Fig. 9 is the functional block diagram for further illustrating object test equipment according to an embodiment of the invention.
As shown in figure 9, object test equipment 90 according to an embodiment of the invention includes memory 901 and processing
Device 902.Specifically, the memory 901 be configured to storage comprising object to be detected anaglyph and
Range image.The processor 902 is coupled to the memory 901, and the processor 902 is configured to:Obtain
Take the anaglyph and range image for including object to be detected;Based on the anaglyph, obtaining has the
The parallax object of one confidence level, and the range image is based on, obtain the distance with the second confidence level
Object;And based on the parallax object with the first confidence level and described with the second confidence level
Apart from object, it is determined that the subject area detected, wherein, first confidence level and second confidence
Degree indicates respectively the parallax object and the confidence level for belonging to same object apart from object.
Figure 10 is the exact functionality for further illustrating object test equipment according to an embodiment of the invention
Block diagram.As shown in Figure 10, object test equipment 91 according to an embodiment of the invention include memory 901,
Processor 902 and input-output equipment 903.
Specifically, the input-output equipment 903 can be received via wired or wireless communication mode and included
The anaglyph and range image of object to be detected are as input, and by the anaglyph and distance of reception
Image is supplied to the memory 901 and the processor 902.The memory 901 is based on storing
Anaglyph and range image of calculation machine programmed instruction and input etc..The processor 902 is coupled to this
Memory 901, the computer program instructions that the memory 901 stores are run by the processor 902
When, the processor 902 operates to the image acquisition component 200, right for including the description of reference picture 1 before
As obtaining widget 300 and subject area obtaining widget 400.Described image obtaining widget 200 is obtained and included
The anaglyph and range image of object to be detected;The object acquisition part 300 is based on the disparity map
Picture, the parallax object with the first confidence level is obtained, and be based on the range image, obtained with the
Two confidence levels apart from object;The subject area obtaining widget 400 has the first confidence level described in being based on
Parallax object and it is described have the second confidence level apart from object, it is determined that the subject area detected.
It is determined that after the subject area detected, the subject area is supplied to described by the processor 901
The subject area detected is presented to described by input-output equipment 903, the input-output equipment 903
The user of object test equipment 91.
Figure 11 is the schematic diagram for illustrating object detection systems according to an embodiment of the invention.
As shown in figure 11, object detection systems 100 according to an embodiment of the invention are caught including the first image
Obtain the image capture apparatus 1003 of device 1002 and second.In one embodiment of the invention, described
One image capture apparatus 1002 is the range image acquisition equipment of such as range laser radar.Second figure
As acquisition equipment 1003 is the anaglyph acquisition equipment of such as binocular camera.Described first image capture dress
Put 1002 and second image capture apparatus 1003 respectively for for the object 1001 in monitoring scene
Image capture process is performed, described first image acquisition equipment 1002 obtains and includes object 1001 to be detected
Range image 1004, second image capture apparatus 1003 obtained comprising object 1001 to be detected
Anaglyph 1005.
Object detection systems 100 according to an embodiment of the invention also include object test equipment 1006.Institute
State range image 1004 and anaglyph 1005 is provided to the object test equipment 1006.It is described right
As detection means 1006 is performed above by reference to the object according to an embodiment of the invention described in Fig. 1 to Fig. 8
Detection method, anaglyph and range image and both putting for object detection based on monitoring scene
Reliability, realize the accurate detection of object in scene.
More than, method for checking object and object according to embodiments of the present invention are described referring to figs. 1 to Figure 11
Detection means.In object test equipment and method for checking object according to an embodiment of the invention, based on tool
There are the parallax object of the first confidence level and the interaction apart from object with the second confidence level, correction
Subject area with different confidence levels, realize the accurate detection of object in scene.That is, root
Overcome according to the object test equipment and method for checking object of embodiments of the invention and be based only on anaglyph
Object detection may cause parallax point discontinuous due to reasons such as shade, low-lights, and be based only on
The object detection of range image may have the noise pixel as caused by the impurity in air due to range image
The defects of, accurately realize the object detection in scene.
It should be noted that in this manual, term " comprising ", "comprising" or its any other
Variant is intended to including for nonexcludability, so that process, method, thing including a series of elements
Product or equipment not only include those key elements, but also the other element including being not expressly set out, or
It is also to include for this process, method, article or the intrinsic key element of equipment.Do not limiting more
In the case of, the key element that is limited by sentence "including a ...", it is not excluded that including the key element
Process, method, other identical element also be present in article or equipment.
Finally, it is to be noted that, a series of above-mentioned processing not only include pressing with order described here
The processing that time series performs, and the place including performing parallel or respectively rather than in chronological order
Reason.
Through the above description of the embodiments, those skilled in the art can be understood that this hair
The bright mode that can add required hardware platform by software is realized, naturally it is also possible to all by hardware come
Implement.Based on such understanding, whole that technical scheme contributes to background technology or
Part can be embodied in the form of software product, and the computer software product can be stored in storage and be situated between
In matter, such as ROM/RAM, magnetic disc, CD, including some instructions are causing a computer to set
Standby (can be personal computer, server, or network equipment etc.) performs each embodiment of the present invention
Or the method described in some parts of embodiment.
The present invention is described in detail above, principle of the specific case used herein to the present invention
And embodiment is set forth, the explanation of above example is only intended to help the method for understanding the present invention
And its core concept;Meanwhile for those of ordinary skill in the art, according to the thought of the present invention,
There will be changes in embodiment and application, in summary, this specification content should not
It is interpreted as limitation of the present invention.
Claims (10)
1. a kind of method for checking object, including:
Obtain the anaglyph and range image for including object to be detected;
Based on the anaglyph, obtain the parallax object with the first confidence level, and based on it is described away from
From image, obtain with the second confidence level apart from object;And
Based on the parallax object with the first confidence level and the distance pair with the second confidence level
As, it is determined that the subject area detected,
Wherein, first confidence level and second confidence level indicate respectively the parallax object and described
Belong to the confidence level of same object apart from object.
2. method for checking object as claimed in claim 1, wherein described be based on the range image, obtain
Take includes with the second confidence level apart from object:
Based on the range image, it is determined that with second confidence level apart from line segment;And
Based on described apart from line segment, obtain with second confidence level apart from object,
Wherein, it is described to include apart from line segment by the first kind in the straight line predetermined distance range away from horizontal direction
The line segment of point composition and/or the line segment being made up of the continuous second class point on oblique line, and based on described first
The coordinate of class point and/or the second class point determines second confidence level.
3. method for checking object as claimed in claim 1 or 2, wherein described have first based on described
The parallax object of confidence level and it is described have the second confidence level apart from object, it is determined that the object detected
Region includes:
By on the plane of delineation of the parallax Object Projection to the range image, the parallax pair of projection is obtained
As;
Match the parallax object of the projection with it is described apart from object;
Compare first confidence level and second confidence level, determine first confidence level and described
High confidence level in two confidence levels;And
Based on the high confidence level, extend or split the parallax object of the projection with it is described in object
One.
4. method for checking object as claimed in claim 3, wherein described have the first confidence based on described
The parallax object of degree and it is described have the second confidence level apart from object, it is determined that the subject area detected
Also include:
It is described after extending or splitting the parallax object of the projection and one in object
The parallax object of projection, apart from the case that object is misaligned, is repeated based on the high confidence level with described
Extension or segmentation, until the parallax object of the projection overlaps with described apart from object,
It is determined that corresponding to coincidence the projection parallax object with it is described apart from the region of object be the inspection
The subject area measured.
5. method for checking object as claimed in claim 1 or 2, in addition to:
Based on the subject area detected, the characteristic information of object in the subject area is updated.
6. a kind of object test equipment, including:
Image acquisition component, for obtaining the anaglyph and range image that include object to be detected;
Object acquisition part, for based on the anaglyph, obtaining the parallax pair with the first confidence level
As, and be based on the range image, obtain with the second confidence level apart from object;
Subject area obtaining widget, for based on the parallax object with the first confidence level and described
With the second confidence level apart from object, it is determined that the subject area detected,
Wherein, first confidence level and second confidence level indicate respectively the parallax object and described
Belong to the confidence level of same object apart from object.
7. object test equipment as claimed in claim 6, wherein the object acquisition part is based on described
Range image, it is determined that with second confidence level apart from line segment;And based on described apart from line segment,
Obtain with second confidence level apart from object,
Wherein, it is described to include apart from line segment by the first kind in the straight line predetermined distance range away from horizontal direction
The line segment of point composition and/or the line segment being made up of the continuous second class point on oblique line, and based on described first
The coordinate of class point and/or the second class point determines second confidence level.
8. object test equipment as claimed in claims 6 or 7, wherein the subject area obtaining widget
Including:
Parallax Object Projection unit, for the image of parallax Object Projection to the range image to be put down
On face, the parallax object of projection is obtained;
Object matching unit, for match the parallax object of the projection with it is described apart from object;
Confidence level comparing unit, for first confidence level and second confidence level, determine institute
State the high confidence level in the first confidence level and second confidence level;And
Subject area corrects unit, for based on the high confidence level, extending or splitting regarding for the projection
Poor object and one in object.
9. object test equipment as claimed in claim 8, wherein extending or splitting regarding for the projection
Poor object with after one in object, the parallax object of the projection with it is described apart from object
In the case of misaligned, the subject area correction unit repeats the extension based on the high confidence level or divided
Cut, until the parallax object of the projection overlaps with described apart from object,
The subject area obtaining widget determine parallax object corresponding to the projection overlapped with it is described away from
Region from object is the subject area detected.
10. object test equipment as claimed in claims 6 or 7, in addition to:
Characteristics of objects updates part, for based on the subject area detected, updating in the subject area
The characteristic information of object.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610320029.7A CN107368837B (en) | 2016-05-13 | 2016-05-13 | Object detection method and object detection device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610320029.7A CN107368837B (en) | 2016-05-13 | 2016-05-13 | Object detection method and object detection device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107368837A true CN107368837A (en) | 2017-11-21 |
CN107368837B CN107368837B (en) | 2020-08-04 |
Family
ID=60303693
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610320029.7A Expired - Fee Related CN107368837B (en) | 2016-05-13 | 2016-05-13 | Object detection method and object detection device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107368837B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103052968A (en) * | 2010-08-03 | 2013-04-17 | 松下电器产业株式会社 | Object detection device, object detection method, and program |
JP2013219579A (en) * | 2012-04-10 | 2013-10-24 | Toshiba Corp | Information processing apparatus, correction method for three-dimensional image, and program |
CN104317391A (en) * | 2014-09-24 | 2015-01-28 | 华中科技大学 | Stereoscopic vision-based three-dimensional palm posture recognition interactive method and system |
CN104574434A (en) * | 2013-10-24 | 2015-04-29 | 株式会社理光 | Object detection method and device |
CN104834887A (en) * | 2014-02-11 | 2015-08-12 | 株式会社理光 | Motion pedestrian representation method, identification method and motion pedestrian identification device |
-
2016
- 2016-05-13 CN CN201610320029.7A patent/CN107368837B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103052968A (en) * | 2010-08-03 | 2013-04-17 | 松下电器产业株式会社 | Object detection device, object detection method, and program |
JP2013219579A (en) * | 2012-04-10 | 2013-10-24 | Toshiba Corp | Information processing apparatus, correction method for three-dimensional image, and program |
CN104574434A (en) * | 2013-10-24 | 2015-04-29 | 株式会社理光 | Object detection method and device |
CN104834887A (en) * | 2014-02-11 | 2015-08-12 | 株式会社理光 | Motion pedestrian representation method, identification method and motion pedestrian identification device |
CN104317391A (en) * | 2014-09-24 | 2015-01-28 | 华中科技大学 | Stereoscopic vision-based three-dimensional palm posture recognition interactive method and system |
Also Published As
Publication number | Publication date |
---|---|
CN107368837B (en) | 2020-08-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10996062B2 (en) | Information processing device, data management device, data management system, method, and program | |
US11763568B2 (en) | Ground plane estimation in a computer vision system | |
CN110322500B (en) | Optimization method and device for instant positioning and map construction, medium and electronic equipment | |
EP2858008A2 (en) | Target detecting method and system | |
CN110587597B (en) | SLAM closed loop detection method and detection system based on laser radar | |
KR101550972B1 (en) | Apparatus and Method for Extracting of Feature Point for Recognize of Obstacle using Laser Scanner | |
US10499038B2 (en) | Method and system for recalibrating sensing devices without familiar targets | |
JP6021689B2 (en) | Vehicle specification measurement processing apparatus, vehicle specification measurement method, and program | |
CN110349212B (en) | Optimization method and device for instant positioning and map construction, medium and electronic equipment | |
JP2021033510A (en) | Driving assistance device | |
Sun et al. | A robust lane detection method for autonomous car-like robot | |
Li et al. | Real-time top-view people counting based on a kinect and nvidia jetson tk1 integrated platform | |
CN112001357B (en) | Target identification detection method and system | |
CN113569812A (en) | Unknown obstacle identification method and device and electronic equipment | |
CN113673288B (en) | Idle parking space detection method and device, computer equipment and storage medium | |
EP3879810A1 (en) | Imaging device | |
CN111144415A (en) | Method for detecting micro pedestrian target | |
CN112819953B (en) | Three-dimensional reconstruction method, network model training method, device and electronic equipment | |
CN107368837A (en) | Method for checking object and object test equipment | |
CN112802112B (en) | Visual positioning method, device, server and storage medium | |
EP3288260A1 (en) | Image processing device, imaging device, equipment control system, equipment, image processing method, and carrier means | |
CN114419564A (en) | Vehicle pose detection method, device, equipment, medium and automatic driving vehicle | |
CN113011212B (en) | Image recognition method and device and vehicle | |
KR20140143918A (en) | Method and Apparatus for Detecting Foregroud Image with Separating Foregroud and Background in Image | |
CN109325962B (en) | Information processing method, device, equipment and computer readable storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200804 |
|
CF01 | Termination of patent right due to non-payment of annual fee |