CN102629317B - Image detection method of farmland weak navigation information and system thereof - Google Patents
Image detection method of farmland weak navigation information and system thereof Download PDFInfo
- Publication number
- CN102629317B CN102629317B CN201210058950.0A CN201210058950A CN102629317B CN 102629317 B CN102629317 B CN 102629317B CN 201210058950 A CN201210058950 A CN 201210058950A CN 102629317 B CN102629317 B CN 102629317B
- Authority
- CN
- China
- Prior art keywords
- image
- detected
- array
- coordinate
- point
- 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.)
- Expired - Fee Related
Links
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an image detection method of farmland weak navigation information and a system thereof, relating to the field of machine vision navigation technology. The method comprises the following steps of: S1, collecting an image to be detected of current farmland; S2, judging whether the image to be detected is a first frame image, if so, carrying out navigation straight line detection of the first frame image, and returning to the step S1, otherwise, carrying out navigation line detection of a non first frame image; S3, carrying out farmland end detection according to a result of the navigation line detection of the non first frame image, judging whether reaches a farmland end or not, if so, ending the detection, and otherwise, returning to the step S1. According to the invention, by targeting a farmland work environment with weak navigation information, utilizing methods of wavelet transform and linear analysis, mutual association of front and rear frames, segmentation Hough transform and the like, in the condition that dividing lines of farmland areas are not obvious, the image identification of a navigation path is realized, and a detection result is accurate, stable and fast.
Description
Technical field
The present invention relates to machine vision navigation technical field, particularly image detecting method and the system of a kind of farmland weak navigation information.
Background technology
Farm working machinery automated navigation system is the important component part of modern agriculture engineering, its in farming, sowing, fertilising, gather in, spray insecticide, there is purposes widely the aspect such as farm environment information acquisition.Along with take the fast development of the precision agriculture that informationization technology is core, the airmanship based on machine vision has obtained general concern with advantages such as its dirigibility, real-time and precision are good.For the obvious operating environment of border region, as farming, harvesting, rice transplanting etc., in image, contain more navigation information, but wait farm work environment for sowing, because area limit is not obvious, seldom, the color of object and non-object, Texture eigenvalue are very approaching for the quantity of information containing in image, and the image recognition of its guidance path is all that a difficult problem is unresolved all the time.
Summary of the invention
(1) technical matters that will solve
The technical problem to be solved in the present invention is: how, in the unconspicuous situation of farmland area limit, realize the image recognition of guidance path.
(2) technical scheme
For solving the problems of the technologies described above, the invention provides the image detecting method of a kind of farmland weak navigation information, said method comprising the steps of:
S1: the image to be detected that gathers current farmland;
S2: judge whether described image to be detected is the first two field picture, if so, carries out the navigation straight-line detection of the first two field picture, returns to step S1, otherwise carry out the navigation straight-line detection of non-the first two field picture;
S3: according to the result of the navigation straight-line detection of described non-the first two field picture, carry out field end detection, judge whether to arrive Tian Duan, if so, detection of end, otherwise, return to step S1.
Preferably, the navigation straight-line detection of described the first two field picture specifically comprises the following steps:
A1: described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row in described image to be detected, then the image after small echo line translation is carried out to inverse transformation, with the image to be detected after obtaining smoothly;
A2: obtain the coordinate of the wave trough position of each row in described image to be detected after level and smooth and be stored in array V, described image to be detected after level and smooth be take upper left side as true origin;
A3: calculate the mean value xva of the x direction coordinate of each point coordinate in described array V, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of the y direction of described image to be detected after level and smooth.
Preferably, in steps A 2, specifically comprise the following steps:
A21: described image to be detected after level and smooth is carried out to projection in y direction to obtain accumulative histogram, calculate the x direction coordinate x of the wave trough position in described accumulative histogram
v;
A22: to the first row in default row in described image to be detected after level and smooth, with x direction coordinate x
vcentered by, the pixel that the first default number is respectively expanded in left and right is scope, finds the coordinate (x of wave trough position
v0, y
v0) and be saved in described array V, to other row in described default row,, centered by the x direction coordinate of the wave trough position that lastrow was obtained, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V;
A23: to other row except described default row in described image to be detected after level and smooth, respectively centered by the mean value of the x direction coordinate of the wave trough position in described default row, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V.
Preferably, further comprising the steps of after steps A 3:
A4: each point coordinate on described navigation straight line is deposited in navigation straight line array L.
5, method as claimed in claim 4, is characterized in that, the navigation straight-line detection of described non-the first two field picture specifically comprises the following steps:
B1: will be more than or equal to preset coordinate (x in array V corresponding to the previous frame image of described image to be detected
0, y
0) the coordinate of y direction value deposit in array Vt, will in navigation straight line array L corresponding to the previous frame image of described image to be detected, be less than preset coordinate (x
0, y
0) the coordinate of y direction value deposit array L in
bin, described image to be detected be take upper left side as true origin;
B2: the mean value x that calculates the x direction coordinate of each point in described array Vt
vta, with point (x
vta, y
0/ 2) be known point, to described array V
tinterior each point carried out the Hough conversion of known point, to obtain fitting a straight line l
t, by described fitting a straight line l
tupper each point coordinate deposits array L in
tin;
B3: respectively with described array L
twith array L
bin centered by each point coordinate, the pixel that each row is respectively expanded the second default number is to the left and right processing region scope;
B4: the image within the scope of the processing region in described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row within the scope of the processing region in described image to be detected, again the image after small echo line translation is carried out to inverse transformation, with the image within the scope of processing region in the image to be detected after obtaining smoothly;
B5: obtain the coordinate of the wave trough position of each row in the image within the scope of the processing region in described image to be detected after level and smooth and be stored in array V;
B6: the mean value x that calculates the x direction coordinate of each point coordinate in described array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
B7: each point coordinate on the navigation straight line of described image to be detected is deposited in navigation straight line array L.
Preferably, the described second default number m calculates by following formula,
m＝tanα×ysize/2
Wherein, the maximum lateral deflection angle of α for allowing, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected.
Preferably, in step S3, specifically judge whether as follows to arrive Tian Duan:
C1: to being greater than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
t;
C2: to being less than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
d;
C3: if meet | x
t-x
d| > 2m, is judged as and arrives Tian Duan, otherwise return to step S1.
The image detecting system that the invention also discloses a kind of farmland weak navigation information, described system comprises:
Image capture module, for gathering the image to be detected in current farmland;
Judgement detection module, for judging whether described image to be detected is the first two field picture, if so, carries out the navigation straight-line detection of the first two field picture, returns to image capture module, otherwise carries out the navigation straight-line detection of non-the first two field picture;
Field end detection module, for carrying out field end detection according to the result of the navigation straight-line detection of described non-the first two field picture, judges whether to arrive Tian Duan, if so, detection of end, otherwise, return to image capture module.
(3) beneficial effect
The present invention is by the farmland operation environment for having weak navigation information, utilize wavelet transformation and linear analysis, front and back frame is interrelated and the method such as segmentation Hough conversion, in the unconspicuous situation of farmland area limit, realized the image recognition of guidance path, and testing result is accurate, stable and quick.
Accompanying drawing explanation
Fig. 1 is the structural representation of the hardware configuration of method of the present invention;
Fig. 2 is according to the process flow diagram of the image detecting method of the farmland weak navigation information of one embodiment of the present invention;
Fig. 3 is the result schematic diagram of navigation straight-line detection of first two field picture of an embodiment of the present invention;
Fig. 4 is the result schematic diagram of navigation straight-line detection of non-first two field picture of an embodiment of the present invention;
Fig. 5 is the result schematic diagram that judges whether to arrive Tian Duan of an embodiment of the present invention.
Embodiment
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.Following examples are used for illustrating the present invention, but are not used for limiting the scope of the invention.
With reference to Fig. 1, the hardware configuration of method of the present invention is the agricultural machinery that has connected video camera, at agricultural machinery when just starting operation, conventionally along the ridge on limit, farmland and the separatrix straight line moving in farmland, in operation process along operation with the separatrix straight line moving on operation ground not.Therefore,, in order to realize self-navigation, need to detect the navigation straight line of these border region.In present embodiment, by video camera, (resolution is 640x480, frame per second was 30 frame/seconds) be arranged on the side, the place ahead of agricultural machinery, from gathering the image of navigation target (ridge and farmland, operation ground and the separatrix on operation ground not) directly over walking target; Height and depression angle that video camera is installed, can obtain on the basis of controlling the needed area limit line length of the normal walking of agricultural machinery, take to photograph near-sighted wild information as far as possible and determine as principle.
Fig. 2 is according to the process flow diagram of the image detecting method of the farmland weak navigation information of one embodiment of the present invention; With reference to Fig. 2, the method for present embodiment comprises the following steps:
S1: the image to be detected that gathers current farmland;
S2: judge whether described image to be detected is the first two field picture, if so, carries out the navigation straight-line detection of the first two field picture, returns to step S1, otherwise carry out the navigation straight-line detection of non-the first two field picture;
S3: according to the result of the navigation straight-line detection of described non-the first two field picture, carry out field end detection, judge whether to arrive Tian Duan, if so, detection of end, otherwise, return to step S1.
Preferably, the navigation straight-line detection of described the first two field picture specifically comprises the following steps:
A1: described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row in described image to be detected, then the image after small echo line translation is carried out to inverse transformation, with the image to be detected after obtaining smoothly;
A2: obtain the coordinate of the wave trough position of each row in described image to be detected after level and smooth and be stored in array V, described image to be detected after level and smooth be take upper left side as true origin;
A3: the mean value x that calculates the x direction coordinate of each point coordinate in described array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of the y direction of described image to be detected after level and smooth.
Preferably, in steps A 2, specifically comprise the following steps:
A21: described image to be detected after level and smooth is carried out to projection in y direction to obtain accumulative histogram, calculate the x direction coordinate x of the wave trough position in described accumulative histogram
v;
A22: to the first row in default row in described image to be detected after level and smooth, with x direction coordinate x
vcentered by, the pixel that the first default number is respectively expanded in left and right is scope, finds the coordinate (x of wave trough position
v0, y
v0) and be saved in described array V, to other row in described default row,, centered by the x direction coordinate of the wave trough position that lastrow was obtained, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V;
A23: to other row except described default row in described image to be detected after level and smooth, respectively centered by the mean value of the x direction coordinate of the wave trough position in described default row, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V.
Preferably, further comprising the steps of after steps A 3:
A4: each point coordinate on described navigation straight line is deposited in navigation straight line array L.
Preferably, the navigation straight-line detection of described non-the first two field picture specifically comprises the following steps:
B1: will be more than or equal to preset coordinate (x in array V corresponding to the previous frame image of described image to be detected
0, y
0) the coordinate of y direction value deposit in array Vt, will in navigation straight line array L corresponding to the previous frame image of described image to be detected, be less than preset coordinate (x
0, y
0) the coordinate of y direction value deposit array L in
bin, described image to be detected be take upper left side as true origin;
B2: calculate described array V
tthe mean value x of the x direction coordinate of interior each point
vta, with point (x
vta, y
0/ 2) be known point, to described array V
tinterior each point carried out the Hough conversion of known point, to obtain fitting a straight line l
t, by described fitting a straight line l
tupper each point coordinate deposits array L in
tin;
B3: respectively with described array L
twith array L
bin centered by each point coordinate, the pixel that each row is respectively expanded the second default number is to the left and right processing region scope;
B4: the image within the scope of the processing region in described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row within the scope of the processing region in described image to be detected, again the image after small echo line translation is carried out to inverse transformation, with the image within the scope of processing region in the image to be detected after obtaining smoothly;
B5: obtain the coordinate of the wave trough position of each row in the image within the scope of the processing region in described image to be detected after level and smooth and be stored in array V;
B6: the mean value x that calculates the x direction coordinate of each point coordinate in described array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
B7: each point coordinate on the navigation straight line of described image to be detected is deposited in navigation straight line array L.
Preferably, the described second default number m calculates by following formula,
m＝tanα×ysize/2
Wherein, the maximum lateral deflection angle of α for allowing, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected.
Preferably, in step S3, specifically judge whether as follows to arrive Tian Duan:
C1: to being greater than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
t;
C2: to being less than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
d;
C3: if meet | x
t-x
d| > 2m, is judged as and arrives Tian Duan, otherwise return to step S1.
Embodiment 1
In the present embodiment, the color image sequence that the navigation straight-line detection of described the first two field picture first arrives for camera acquisition, find out again color component maximum in entire image and this component image is carried out to the line translation of one dimension Daubechies small echo, remove high frequency noise, realize picture smooth treatment, then calculate the vertical accumulative histogram of level and smooth rear image and determine wave trough position, with trough, put beginning, to lower and on analyze line by line linear feature, find candidate point, finally complete the detection of navigation straight line.With reference to Fig. 3, in the present embodiment, the concrete steps of the navigation straight-line detection of described the first two field picture are as follows:
(1), utilize the Daubechies small echo of wavelet coefficient N=8 to convert line by line entire image, remove after the high fdrequency component in each row, carry out inverse transformation, so carry out obtaining for 3 times the image after level and smooth.
(2), the view data after level and smooth is carried out to projection (being y direction projection) in vertical direction to obtain accumulative histogram, calculate the coordinate of wave trough position in accumulative histogram and be designated as x
v, definition is for depositing the array V of each row candidate point, afterwards from below image, and the front 10 row pixels of bottom-up scanning.When scanning the first row pixel, with x
vfor mid point, 5 pixels are respectively expanded as scope in left and right, find wave trough position and be denoted as (x within the scope of this
v0, y
v0) and deposit in array V.While scanning the second row pixel, with x
v0centered by, same left and right is respectively expanded the wave trough position that 5 pixels find the second row as scope and is denoted as (x
v1, y
v1) and deposit in array V.8 row (as shown in Figure 2) by that analogy afterwards.
(3), since the 10th row, each mean value of all usining its front 10 row wave trough position is as center, left and right is respectively expanded 5 pixels and is found current line trough until image apex searches out after trough at every turn as sweep limit, all its position coordinates is deposited in array V.
(4), in calculating array V, the mean value of the x direction coordinate of each point coordinate is designated as x
va, with point (x
va, the quick Hough conversion of ysize/2) for known point, array V being carried out to known point obtains navigation straight line.The coordinate of the each point of statistics navigation straight line also deposits in array L, and wherein in L, coordinate number is identical with the number of pixels of picture altitude ysize.
In the present embodiment, since the 2nd two field picture, later each two field picture all and its former frame carry out associatedly, utilize the candidate point group of previous frame to carry out segmentation Hough conversion, the straight line that converts acquisition according to Hough redefines the processing region of each row and carries out wavelet Smoothing processing.Wherein, the image that carries out wavelet Smoothing is similarly the component image of maximum color in coloured image.Afterwards, the linear feature of image line inner analysis after smoothly, the candidate point group of searching present frame, completes the detection of navigation straight line.With reference to Fig. 4, the concrete steps of the navigation straight-line detection of described non-the first two field picture are as follows.
(1), each row candidate point group data in the previous frame image top ysize/4 length of depositing in array V is deposited in array Vt, the mean value that calculates the x direction coordinate of each point in Vt is designated as x
vta, with point (x
vta, ysize/8) being known point, the Hough that Vt was carried out to known point converts and obtains fitting a straight line l
t, afterwards by l
tupper each point coordinate deposits array L in
tin.
(2), deposit the data point in presentation video below 3ysize/4 length in array L in array L
bin.
(3), for image top ysize/4 length, with array L
tcentered by middle each point data, each row is expanded respectively m pixel wide (m=tan α * ysize/2, wherein, the maximum lateral deflection angle that α allows while being operation, and in the present embodiment, value is 3 °) to the left and right, and dwindling lateral processes regional extent is l
ti-m to l
ti+ m (as shown in phantom in Figure 4; Wherein, l
tirepresent array L
tin the x direction coordinate of each data) and wavelet Smoothing processing (this processing mode is identical with the mode of the navigation straight-line detection of described the first two field picture) is carried out in this region.Afterwards, calculate the trough information of below, ysize/8 place in y direction (center of ysize/4 length) 10 row, from ysize/8, start upwards to find until image apex (this searching mode is identical with the mode of the navigation straight-line detection of described the first two field picture).In like manner, for the latter half in ysize/4 region, first obtain the wave trough position information of top, ysize/8 place 10 row in y direction, from ysize/8, start to find until image ysize/4 place downwards afterwards.In search procedure, after finding corresponding candidate point, its correspondence is deposited in array V.
(4) for image below 3ysize/4 length, with array L
bcentered by middle each point data, each row is expanded respectively m pixel wide equally to the left and right, and dwindling lateral processes regional extent is l
bi-m to l
bi+ m (as shown in phantom in Figure 4; Wherein, l
birepresent array L
bin the x direction coordinate of each data) and wavelet Smoothing processing is carried out in this region.The trough information of 5ysize/8 place in computed image y direction (center of image below 3ysize/4 length) below 10 row, afterwards from 5ysize/8 start to utilize the method described in the step (3) of navigation straight-line detection of similar described the first two field picture upwards searching until image ysize/4 place.In like manner, for the latter half in 3ysize/4 region, adopt similar approach to complete the searching of each row candidate point.After finding corresponding candidate point, its correspondence is deposited in array V.
(5) the Hough conversion that, utilizes the method in the step (4) of navigation straight-line detection of similar described the first two field picture to carry out known point to array V obtains the navigation straight line of present frame and each coordinate of its each point is deposited in array L.
In the present embodiment, since the 2nd two field picture, need to consider whether Work machine arrives Tian Duan.Due to field, end place there will not be area limit line, therefore can utilize this feature to complete the detection of Tian Duan, with reference to Fig. 5, judges whether as follows to arrive Tian Duan in the present embodiment:
Complete after the navigation straight-line detection of described non-the first two field picture, check in current frame image in y direction the ysize/4 place candidate point group data of each 10 row up and down.Read the data of each 10 row up and down of ysize/4 place in array V.Calculate the mean value of its x direction coordinate, and be designated as respectively x
tand x
d.If | x
t-x
d| > 2m, think and reached Tian Duan, stop detecting (as shown in Figure 5), otherwise proceed the collection of image.
The image detecting system that the invention also discloses a kind of farmland weak navigation information, described system comprises:
Image capture module, for gathering the image to be detected in current farmland;
Judgement detection module, for judging whether described image to be detected is the first two field picture, if so, carries out the navigation straight-line detection of the first two field picture, returns to image capture module, otherwise carries out the navigation straight-line detection of non-the first two field picture;
Field end detection module, for carrying out field end detection according to the result of the navigation straight-line detection of described non-the first two field picture, judges whether to arrive Tian Duan, if so, detection of end, otherwise, return to image capture module.
Above embodiment is only for illustrating the present invention; and be not limitation of the present invention; the those of ordinary skill in relevant technologies field; without departing from the spirit and scope of the present invention; can also make a variety of changes and modification; therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.
Claims (4)
1. an image detecting method for farmland weak navigation information, is characterized in that, said method comprising the steps of:
S1: the image to be detected that gathers current farmland;
S2: judge whether described image to be detected is the first two field picture, if so, carries out the navigation straight-line detection of the first two field picture, returns to step S1, otherwise carry out the navigation straight-line detection of non-the first two field picture;
S3: according to the result of the navigation straight-line detection of described non-the first two field picture, carry out field end detection, judge whether to arrive Tian Duan, if so, detection of end, otherwise, return to step S1;
The navigation straight-line detection of described the first two field picture specifically comprises the following steps:
A1: described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row in described image to be detected, then the image after small echo line translation is carried out to inverse transformation, with the image to be detected after obtaining smoothly,
A2: obtain the coordinate of the wave trough position of each row in described image to be detected after level and smooth and be stored in array V, described image to be detected after level and smooth be take upper left side as true origin,
A3: the mean value x that calculates the x direction coordinate of each point coordinate in described array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of the y direction of described image to be detected after level and smooth;
Described navigation straight-line detection of carrying out non-the first two field picture, specifically comprises the following steps:
B1: will be more than or equal to preset coordinate (x in array V corresponding to the previous frame image of described image to be detected
0, y
0) the coordinate of y direction value deposit array V in
tin, will in navigation straight line array L corresponding to the previous frame image of described image to be detected, be less than preset coordinate (x
0, y
0) the coordinate of y direction value deposit array L in
bin, described image to be detected be take upper left side as true origin;
B2: calculate described array V
tthe mean value x of the x direction coordinate of interior each point
vta, with point (x
vta, y
0/ 2) be known point, to described array V
tinterior each point carried out the Hough conversion of known point, to obtain fitting a straight line l
t, by described fitting a straight line l
tupper each point coordinate deposits array L in
tin;
B3: respectively with described array L
twith array L
bin centered by each point coordinate, the pixel that each row is respectively expanded the second default number is to the left and right processing region scope;
B4: the image within the scope of the processing region in described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row within the scope of the processing region in described image to be detected, again the image after small echo line translation is carried out to inverse transformation, with the image within the scope of processing region in the image to be detected after obtaining smoothly;
B5: obtain the coordinate of the wave trough position of each row in the image within the scope of the processing region in described image to be detected after level and smooth and be stored in array V;
B6: the mean value x that calculates the x direction coordinate of each point coordinate in described array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
B7: each point coordinate on the navigation straight line of described image to be detected is deposited in navigation straight line array L;
The described second default number m calculates by following formula,
m＝tanα×ysize/2
Wherein, the maximum lateral deflection angle of α for allowing, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
The described result according to the navigation straight-line detection of described non-the first two field picture is carried out field end detection, judges whether to arrive Tian Duan, specifically comprises the following steps:
C1: to being greater than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
t;
C2: to being less than preset coordinate (x in array V corresponding to the image within the scope of the processing region in described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
d;
C3: if meet | x
t-x
d| > 2m, is judged as and arrives Tian Duan, otherwise return to step S1.
2. the method for claim 1, is characterized in that, in steps A 2, specifically comprises the following steps:
A21: described image to be detected after level and smooth is carried out to projection in y direction to obtain accumulative histogram, calculate the x direction coordinate x of the wave trough position in described accumulative histogram
v;
A22: to the first row in default row in described image to be detected after level and smooth, with x direction coordinate x
vcentered by, the pixel that the first default number is respectively expanded in left and right is scope, finds the coordinate (x of the wave trough position of described the first row
v0, y
v0) and be saved in described array V, to other row in described default row,, centered by the x direction coordinate of the wave trough position that lastrow was obtained, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V;
A23: to other row except described default row in described image to be detected after level and smooth, respectively centered by the mean value of the x direction coordinate of the wave trough position in described default row, the pixel that the described first default number is respectively expanded in left and right is scope, finds the coordinate of wave trough position and is saved in described array V.
3. the method for claim 1, is characterized in that, further comprising the steps of after steps A 3:
A4: each point coordinate on described navigation straight line is deposited in navigation straight line array L.
4. an image detecting system for farmland weak navigation information, is characterized in that, described system comprises:
Image capture module: for gathering the image to be detected in current farmland;
Judgement detection module comprises: the first two field picture judge module, the first two field picture navigation straight-line detection module, non-the first two field picture navigation straight-line detection module; Described the first two field picture judge module, for judging whether described image to be detected is the first two field picture; Described the first two field picture navigation straight-line detection module, while being the first two field picture for judging described image to be detected when described the first two field picture judge module, carries out the navigation straight-line detection of the first two field picture, returns to image capture module; Described non-the first two field picture navigation straight-line detection module, while being not the first two field picture for judging described image to be detected when described the first two field picture judge module, carries out the navigation straight-line detection of non-the first two field picture;
Field end detection module, for carrying out field end detection according to the result of the navigation straight-line detection of described non-the first two field picture, judges whether to arrive Tian Duan, if so, detection of end, otherwise, return to image capture module;
The navigation straight-line detection module of described the first two field picture, comprising:
A1 module: for described image to be detected is carried out to small echo line translation, to remove the high fdrequency component of each row in described image to be detected, then the image after small echo line translation is carried out to inverse transformation, with the image to be detected after obtaining smoothly,
A2 module: for obtain described each row of image to be detected after level and smooth wave trough position coordinate and be stored in array V, described image to be detected after level and smooth be take upper left side as true origin,
A3 module: for calculating the mean value x of the x direction coordinate of described each point coordinate of array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of the y direction of described image to be detected after level and smooth;
The navigation straight-line detection module of described non-the first two field picture, comprising:
B1 module: for array V corresponding to the previous frame image of described image to be detected is more than or equal to preset coordinate (x
0, y
0) the coordinate of y direction value deposit array V in
tin, will in navigation straight line array L corresponding to the previous frame image of described image to be detected, be less than preset coordinate (x
0, y
0) the coordinate of y direction value deposit array L in
bin, described image to be detected be take upper left side as true origin;
B2 module: for calculating described array V
tthe mean value x of the x direction coordinate of interior each point
vta, with point (x
vta, y
0/ 2) be known point, to described array V
tinterior each point carried out the Hough conversion of known point, to obtain fitting a straight line l
t, by described fitting a straight line l
tupper each point coordinate deposits array L in
tin;
B3 module: for respectively with described array L
twith array L
bin centered by each point coordinate, the pixel that each row is respectively expanded the second default number is to the left and right processing region scope;
B4 module: carry out small echo line translation for the image within the scope of the processing region of described image to be detected, to remove the high fdrequency component of each row within the scope of the processing region in described image to be detected, again the image after small echo line translation is carried out to inverse transformation, with the image within the scope of processing region in the image to be detected after obtaining smoothly;
B5 module: for obtaining the coordinate of the wave trough position of each row in the image within the scope of the processing region of described image to be detected after level and smooth and being stored in array V;
B6 module: for calculating the mean value x of the x direction coordinate of described each point coordinate of array V
va, with point (x
va, ysize/2) be known point, the each point in described array V was carried out to the Hough conversion of known point, to obtain navigation straight line, wherein, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
B7 module: for each point coordinate on the navigation straight line of described image to be detected being deposited in to navigation straight line array L;
The described second default number m calculates by following formula,
m＝tanα×ysize/2
Wherein, the maximum lateral deflection angle of α for allowing, ysize is the number of pixels of y direction in the image within the scope of the processing region in described image to be detected;
Described field end detection module specifically comprises:
C1 module: for being greater than preset coordinate (x in array V corresponding to the image within the scope of the processing region of described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
t;
C2 module: for being less than preset coordinate (x in array V corresponding to the image within the scope of the processing region of described image to be detected after level and smooth
0, y
0) the coordinate of default number of y direction value calculate the mean value x of x direction
d;
C3 module: if for meeting | x
t-x
d| > 2m, is judged as and arrives Tian Duan, otherwise return to image capture module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210058950.0A CN102629317B (en) | 2012-03-07 | 2012-03-07 | Image detection method of farmland weak navigation information and system thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210058950.0A CN102629317B (en) | 2012-03-07 | 2012-03-07 | Image detection method of farmland weak navigation information and system thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102629317A CN102629317A (en) | 2012-08-08 |
| CN102629317B true CN102629317B (en) | 2014-08-20 |
Family
ID=46587575
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210058950.0A Expired - Fee Related CN102629317B (en) | 2012-03-07 | 2012-03-07 | Image detection method of farmland weak navigation information and system thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102629317B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105116911B (en) * | 2015-07-20 | 2017-07-21 | 广州极飞科技有限公司 | Unmanned plane spray method |
| CN111724340A (en) * | 2020-05-09 | 2020-09-29 | 北京农业智能装备技术研究中心 | Farmland margin line visual detection method and system |
-
2012
- 2012-03-07 CN CN201210058950.0A patent/CN102629317B/en not_active Expired - Fee Related
Non-Patent Citations (9)
| Title |
|---|
| Automation》.2010,第2卷 * |
| Liu Changqing等.Study on the Image Processing Algorithm for Detecting the Seed-Sowing Performance.《2010 International Conference on Digital Manufacturing & Automation》.2010,第2卷 |
| Liu Changqing等.Study on the Image Processing Algorithm for Detecting the Seed-Sowing Performance.《2010 International Conference on Digital Manufacturing & * |
| 免耕覆盖地秸秆行茬导航路径的图像检测;王晓燕等;《农业机械学报》;20090625;第40卷(第6期);第158-163页 * |
| 基于机器视觉的导航实验平台的研究;张磊等;《中国农业机械学会2006年学术年会论文集》;20061101;第480-484页 * |
| 基于机器视觉的耕作机器人行走目标直线检测;赵颖等;《农业机械学报》;20060425;第37卷(第4期);第83-86页 * |
| 张磊等.基于机器视觉的导航实验平台的研究.《中国农业机械学会2006年学术年会论文集》.2006, |
| 王晓燕等.免耕覆盖地秸秆行茬导航路径的图像检测.《农业机械学报》.2009,第40卷(第6期), |
| 赵颖等.基于机器视觉的耕作机器人行走目标直线检测.《农业机械学报》.2006,第37卷(第4期), |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102629317A (en) | 2012-08-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Meng et al. | Development of agricultural implement system based on machine vision and fuzzy control | |
| CN103198477B (en) | Apple fruitlet bagging robot visual positioning method | |
| CN106125087A (en) | Dancing Robot indoor based on laser radar pedestrian tracting method | |
| Zhong et al. | A method for litchi picking points calculation in natural environment based on main fruit bearing branch detection | |
| WO2021226900A1 (en) | Cotton crop row detection method and apparatus based on computer vision, and storage medium | |
| CN103186773B (en) | A kind of early stage seedling field line recognizer based on one dimension Hough conversion and expert system | |
| CN110334678A (en) | A kind of pedestrian detection method of view-based access control model fusion | |
| CN104036524A (en) | Fast target tracking method with improved SIFT algorithm | |
| CN102368158A (en) | Navigation positioning method of orchard machine | |
| CN105307115A (en) | Distributed vision positioning system and method based on action robot | |
| CN108133471B (en) | Robot navigation path extraction method and device based on artificial bee colony algorithm | |
| CN109001757A (en) | A kind of parking space intelligent detection method based on 2D laser radar | |
| CN107945215A (en) | High-precision infrared image tracker and a kind of target fast tracking method | |
| CN110555867B (en) | Multi-target object tracking method integrating object capturing and identifying technology | |
| CN105307116A (en) | Distributed vision positioning system and method based on mobile robot | |
| Li et al. | Image detection and verification of visual navigation route during cotton field management period | |
| CN109764809A (en) | A method of fruit tree canopy volume is calculated based on two-dimensional laser sensor in real time | |
| CN103577833A (en) | Abnormal intrusion detection method based on motion template | |
| Ma et al. | A method of multiple lane detection based on constraints of lane information | |
| CN102629317B (en) | Image detection method of farmland weak navigation information and system thereof | |
| Tu et al. | An efficient crop row detection method for agriculture robots | |
| CN102640622B (en) | Cotton picker navigation information image detection method and system | |
| CN112197705A (en) | Fruit positioning method based on vision and laser ranging | |
| Liu et al. | Extracting visual navigation line between pineapple field rows based on an enhanced YOLOv5 | |
| CN114689038A (en) | Fruit detection positioning and orchard map construction method based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140820 Termination date: 20160307 |