CN104574312A - Method and device of calculating center of circle for target image - Google Patents
Method and device of calculating center of circle for target image Download PDFInfo
- Publication number
- CN104574312A CN104574312A CN201510005605.4A CN201510005605A CN104574312A CN 104574312 A CN104574312 A CN 104574312A CN 201510005605 A CN201510005605 A CN 201510005605A CN 104574312 A CN104574312 A CN 104574312A
- Authority
- CN
- China
- Prior art keywords
- image
- circle
- target image
- pixel
- absolute value
- 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.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a method of calculating a center of a circle for a target image. The method includes the following steps that an original image is read and a gaussian filter process is conducted on the original image, the original image comprises a target image area, and the target image comprises at least one circular image area; edge extraction is conducted on the original image processed by the gaussian filter process and the maximum range of the target image area in the original image is determined, and edge of each circular image is determined in the maximum range of the target image area; coordinates of the center of a circle of the same circular image are calculated and a difference value between the two coordinates of the centers of the circles is checked through an ellipse fitting method and a centroid method based on the corresponding coordinates of the edge of the circle image. The invention further provides a device for calculating the center of the circle for the target image. The coordinates of the center of the circle of each target image are calculated to acquire characteristic value information of an image coordinate system of a calibrating plate in the calibrating process, the image coordinates of the target image are determined, and a congruent relationship between the world coordinates and the image coordinates is established according to the image coordinates.
Description
Technical field
The present invention relates to computer vision field, particularly relate to the method and apparatus that target image asks the center of circle.
Background technology
Automobile four-wheel location technology be in the past using sensor as measure Data Source, but along with theory of stereo vision ripe gradually, 3D four-wheel aligner system is arisen at the historic moment.Between image coordinate and world coordinates, set up particular kind of relationship, in three dimensions, using image as analysis foundation, each angle information of four-wheel can be obtained.Can complete a whole set of four-wheel aligner work by camera and target, the weight that four-wheel aligner is equipped and cost decline to a great extent, and maintenance cost also reduces relatively.
In machine vision, image measurement, determine three-dimensional geometry position and its mutual relationship in the picture between corresponding point of certain point of space object surface, set up the geometric model of camera imaging.This geometric model except there being round spot, also has chessboard pattern.What chessboard calibration plate adopted is that measured angular dot information is to determine image coordinate system, in contrast, the coordinate system that the scaling board justifying shape of spot is demarcated out, can not have influence on target image because shooting angle is different, therefore the data obtained is more stable, and accuracy is also higher.In numerous graphical analysis and disposal system, obtain the center of circle quickly and accurately and be widely used on target identification and location, in the system of three-dimensionalreconstruction, obtaining the scaling board center of circle accurately has vital impact to whole three-dimensionalreconstruction precision.
The algorithm in the target image center of circle of scaling board of asking common at present has: centroid method, Hough transform method, Gauss curved fitting process etc.Wherein, centroid method requires that gradation of image distribution is relatively more even, otherwise can produce comparatively big error; Hough transform rule needs each frontier point pointwise ballot, record, and computing time is grown and takies larger calculator memory, is restricted in actual applications; Gauss curved the rule of fitting utilizes the intensity profile of circle spot be similar to Gauss model principle and carry out surface fitting, and its positioning precision is high, if but the round spot area of process is comparatively large, then and operand also will increase, and is not therefore applicable to actual detection very much.
Summary of the invention
Fundamental purpose of the present invention is to provide a kind of target image to ask the method and apparatus in the center of circle, aims to provide a kind of object of central coordinate of circle of round spot of easy accurate calculation target image.
For achieving the above object, the invention provides a kind of method that target image asks the center of circle, described target image comprises at least one circle spot image-region, comprising:
Read original image and carry out gaussian filtering process to described original image, described original image comprises target image region;
Edge extracting is carried out to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
In the maximum magnitude in described target image region, determine the edge of each described round spot image;
The coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval;
When the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserve described central coordinate of circle.
Preferably, described edge extracting is carried out to the described original image after described gaussian filtering process, determines that the maximum magnitude in target image region described in described original image comprises:
In described original image each pixel eight neighborhood in, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
Carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
Preferably, described in the maximum magnitude in described target image region, determine that the edge of each described round spot image comprises:
The mean flow rate of computed image;
In the maximum magnitude in described target image region, according to the mean flow rate of described original image, determine the edge of each described round spot image.
For achieving the above object, the present invention also provides a kind of target image to ask the device in the center of circle, and described target image comprises at least one circle spot image, comprising:
Pretreatment module, for reading original image and carrying out gaussian filtering process to described original image, described original image comprises target image region;
Target image edge determination module, for carrying out edge extracting to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
Circle spot image border determination module, in the maximum magnitude in described target image region, determines the edge of each described round spot image;
Central coordinate of circle calculation check module, for the coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval;
Central coordinate of circle preserves module, for when the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserves described central coordinate of circle.
Preferably, described target image edge determination module comprises:
Neighborhood gray-scale value difference computational unit, in eight neighborhood for each pixel in described original image, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
Screening identify unit, for carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
Preferably, described round spot image border determination module comprises:
Average luminance computing unit, for the mean flow rate of computed image;
Circle spot image border determining unit, in the maximum magnitude in described target image region, according to the mean flow rate of described original image, determines the edge of each described round spot image.
The present invention first by carrying out gaussian filtering process to original image, to eliminate the impact of the Gaussian noise in original image for computation process; And then pass through the eight neighborhood gray scale difference value computing method of image, edge extracting is carried out to original image, thus target image is separated from the background of original image; And then from described target image, determine the edge of each described round spot image again; Last coordinate corresponding according to the edge of described round spot image again, is calculated respectively by ellipse fitting method and gravity model appoach and obtains the central coordinate of circle of same described round spot image.Fast target image can be separated from the background of original image in the scope of acceptable error by the eight neighborhood gray scale difference value computing method of image.And then carry out iterative computation according to the mean flow rate of image, thus obtain the optimum segmentation threshold value of justifying spot image more accurately further, it is last that by ellipse fitting method and gravity model appoach, the center of circle to circle spot image solves, accurately to filter out satisfactory central coordinate of circle again.
Accompanying drawing explanation
Fig. 1 is the exemplary plot of pretreatment image of the present invention;
Fig. 2 is the schematic flow sheet that target image of the present invention asks method first embodiment in the center of circle;
Fig. 3 is the schematic flow sheet that target image of the present invention asks method second embodiment in the center of circle;
Fig. 4 is the exemplary plot of the pixel eight neighborhood of image;
Fig. 5 is the schematic flow sheet that target image of the present invention asks method the 3rd embodiment in the center of circle;
Fig. 6 is the exemplary plot of the round spot edge that the present invention extracts;
Fig. 7 is the high-level schematic functional block diagram that target image of the present invention asks device first embodiment in the center of circle;
Fig. 8 is the high-level schematic functional block diagram that target image of the present invention asks device second embodiment in the center of circle;
Fig. 9 is the high-level schematic functional block diagram that target image of the present invention asks device the 3rd embodiment in the center of circle.
The realization of the object of the invention, functional characteristics and advantage will in conjunction with the embodiments, are described further with reference to accompanying drawing.
Embodiment
Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
The present invention is according to perspective projection principle, the image obtained after pin-hole imaging for the circle in object plane can think ellipse (when not considering distortion), its elliptical center then can think the point that the center of circle obtains after mapping, by obtaining elliptical center coordinate to set up the relation between camera review location of pixels and scene point location.
Camera calibration is the important step of machine vision metrology, based on camera perspective projection model, by demarcating the relation set up between camera review location of pixels and scene point location.The feature point for calibration that camera calibration need provide by scaling board, solves the model parameter of video camera according to the image coordinate of known calibration unique point and world coordinates.The world coordinates of feature point for calibration is determined when making by scaling board, and image coordinate then needs to ask for by processing the target image of shooting and analyze.At present, more 2D plane reference plate is used to have chessboard calibration plate, grid scaling board and dot matrixes scaling board etc. in camera calibration.Its feature point for calibration of flat circle lattice array scaling board is round dot center.The rectangular array composition of the equidistant distribution that the pattern of flat circle lattice array scaling board forms primarily of multiple round dot, the method that plated film on pmma substrate can be adopted to portray pattern makes, and its precision can reach ± 0.001mm.The present invention is specifically applied to automobile 3D four-wheel aligner technology with flat circle lattice array target and illustrates, but thought of the present invention does not limit the field being applied to dot matrixes target and its application.
With reference to the exemplary plot that Fig. 1, Fig. 1 are pretreatment image of the present invention.Described pretreatment image 1 comprises background image 2 and scaling board image 3, and wherein, scaling board image 3 comprises multiple elliptical target image 4.In described pretreatment image 1, the gray-scale value corresponding due to the pixel of the two between background image 2 from scaling board image 3 is different, and the gray-scale value that between target image 4 from scaling board image 3, the pixel of the two is corresponding is also different, thus three can be distinguished by the method for edge extracting.Therefore, before the central coordinate of circle calculating each ellipse in oval target image 4, need first scaling board image 3 to be separated from background image 2.
Reference Fig. 2, Fig. 2 are the schematic flow sheet that target image of the present invention asks method first embodiment in the center of circle.In the present embodiment, described target image asks the method in the center of circle to comprise:
Step S10, read original image and carry out gaussian filtering process to described original image, described original image comprises target image region;
DIB (Device-Independent-Bitmap is used in the coordinate system of image acquiescence, device independent bit) class captured bitmap file when reading camera calibration, to ensure that the bitmap graphics created by certain application program can be loaded by other application program or show.Wherein, DIB can display bitmap is intrinsic in different machines or system color.For accelerating the processing speed of pre-set image, gray processing process can be carried out according to the form of bitmap pixels and obtaining gray level image.Then, gaussian filtering method is adopted to the smoothing process of pre-set image to improve the image quality of image and to obtain pending image.
Gaussian filtering is a kind of linear smoothing filtering, is applicable to eliminate Gaussian noise.The specific operation process of gaussian filtering is: by each pixel in a template scan image, and the weighted mean gray-scale value of pixel goes the value of alternate template central pixel point in the neighborhood determined by template, also namely the value of each pixel is obtained after weighted mean by other pixel values in itself and neighborhood in view picture figure.
Step S20, carries out edge extracting to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
Described original image again after gaussian filtering process, then carries out edge extracting, to be separated from the background of described original image by scaling board image.Have a lot to the method for Image Segmentation Using, the present invention is under the prerequisite considering computing velocity and degree of accuracy, and in conjunction with follow-up calculation procedure and method, in this step, preferably by rim detection, the method for image border is extracted for the dividing method of scaling board image.
Iamge Segmentation refers to the region according to features such as gray scale, color, texture and shapes, image being divided into some mutual not crossovers, and makes these features in the same area, present similarity, and between zones of different, present obvious otherness.On the boundary line that the edge of image refers to two zoness of different in image, the set of continuous print pixel, is the reflection of image local feature uncontinuity, embodies the sudden change of the picture characteristics such as gray scale, color, texture.Under normal circumstances, based on the rim detection that the image partition method at edge is based on gray-scale value.
Step S30, in the maximum magnitude in described target image region, determines the edge of each described round spot image;
By the calculating of step S20, determine the maximum magnitude of target image, in the maximum magnitude of target image, continue the edge determining each described round spot image.Determine that the method at the edge of circle spot image has a lot, such as comparatively common Roberts Cross operator, Prewitt operator, Sobel operator, Canny operator, Kirsch operator etc., often kind of method has its merits and demerits, when carrying out justifying spot identification, select compatibly round spot center identification method extremely important, it directly affects the precision of measurement.
Step S40, the coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval;
The coordinate corresponding according to the edge of each round spot image determined in step S30, in the present embodiment, calculates the central coordinate of circle of same round spot image respectively by ellipse fitting method and gravity model appoach.After asking central coordinate of circle by gravity model appoach and ellipse fitting method to same round spot, whether the difference between the coordinate in respectively described two centers of circle of corresponding verification is within tolerance interval.Such as, the central coordinate of circle that gravity model appoach calculates is (X
1, Y
1)=(3.1,5.7), the central coordinate of circle that ellipse fitting method calculates is (X
2, Y
2)=(6.5,9.7), then coordinate 1 (X
1, Y
1) and coordinate 2 (X
2, Y
2) between the absolute value of difference be respectively | X
1-X
2|=3.4, | Y
1-Y
2|=4.For ensureing the accuracy of result of calculation further, the present invention adopts the center of circle of two kinds of methods to circle spot to solve to limit the difference that both obtain the center of circle, if the difference difference of the two is too large, then this center of circle is not the correct center of circle, if within the acceptable range, then illustrate that this center of circle is the center of circle of needs.
Curve is one of the most basic problem in computer vision, its objective is to extract special characteristic information for subsequent treatment from image.Conventional ellipse fitting method mainly contains three classes, and a class is the ellipse fitting method based on Hough transform, and another kind of is method based on not bending moment, and the 3rd class is then the method based on least square method.In these methods, the method based on least square method is more suitable for the handling object model of various complexity, and can provide estimating about error of fitting very intuitively, thus reaches very high fitting precision.
The basic ideas of ellipse fitting method are: for group sample point of in given plane, find an ellipse, make it as far as possible near these sample points.That is arrive, be that model carries out matching with elliptic equation by group data of in image, make a certain elliptic equation meet these data as far as possible, and obtain the parameters of this elliptic equation.Finally determine that namely best oval center is the center of circle of the round spot that will determine.
Least square method basic thought considers that data are subject to the impact of random noise and then pursue minimizing of global error.For ellipse fitting, be exactly first suppose elliptic parameter, obtain each distance sum that match point arrives this ellipse for the treatment of, namely to the error that hypothesis is oval, obtain and make this and minimum parameter.Least square method is mainly used in the matching of curve, and it finds the optimal function coupling of data by the quadratic sum of minimum error.Utilize least square method can try to achieve unknown data easily, and between the data that these are tried to achieve and real data, the quadratic sum of error is minimum.。
Ellipse formula general expression: Ax
2+ By
2+ Cxy+Dx+Ey+F=0
Least square method formula (y=ax+b form):
Wherein, N represents N group (x, y) coordinate,
for the average of y value in all N group coordinates,
for the average of x value in all N group coordinates.
Gravity model appoach can regard the weighting type heart method using the gray-scale value of each pixel as weights as.In the target image, the center of gravity C (C of its image
x, C
y) computing formula be:
Wherein, C
xfor the x coordinate of center of gravity; C
yfor the y coordinate of center of gravity; D
ixthe x coordinate of i-th pixel in target image; D
iythe y coordinate of i-th pixel in target image; V
ifor the gray-scale value of i-th pixel in target image.
Gravity model appoach for background gray levels less and target gray value is higher and the gray scale of targeted graphical be respectively parabola or Gauss curved the center of circle location, higher positioning precision can be obtained, be applicable to the inventive method application automobile 3D four-wheel aligner technical field.
Step S50, when the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserves described central coordinate of circle.
In the present embodiment preferably 0.6 pixel as the standard of the absolute value of difference corresponding to measurement two kinds of methods.Such as, as | X
1-X
2| <=0.6 and | Y
1-Y
2| during <=0.6, then the result accepting to draw is correct central coordinate of circle and preserves.
In the present embodiment, first target image is separated from pre-set image to reduce the scope of searching calculating further; Next, then from target image, determine the edge of each circle spot; Last set of coordinates corresponding according to the round spot edge determined again, calculates the center of circle of this circle spot respectively by gravity model appoach and ellipse fitting method.For promoting speed and the precision of measuring and calculating further, in conjunction with computing velocity and accuracy requirement in the present embodiment, the computing method adapted are selected in the different measuring and calculating stages, in the hope of reaching the balance in computing velocity and precision, thus provide a kind of and calculate more fast and meet the combination computing method asking the round spot center of circle of accuracy requirement.
In addition, for ensureing the accuracy of result of calculation further, the present invention obtains the difference in the center of circle by asking the central coordinate of circle of round spot to two kinds of methods with both restrictions, if the difference difference of the two is too large, then illustrate that this center of circle is not the correct center of circle, and then calculate the central coordinate of circle of next circle spot; If within the acceptable range, then illustrate that this center of circle is the center of circle of needs, preserve the coordinate in the described center of circle.The image in different application field asks the center of circle, its requirement for computing method is all not quite similar, simultaneously, different computing method ask its effect calculated of the center of circle also all different for the image in different technologies field, but various method is mainly weighed from the speed calculated and precision two aspects for the requirement of the result calculated.In the present embodiment, according to the feature of place of the present invention application, and combine the speed of calculating and considering of precision two aspect, present invention improves over the method in original computed image center of circle and define a set of method being applicable to comprise the technical applications computed image center of circle that automobile 3D four-wheel detects.
Further, reference Fig. 3, Fig. 3 is the schematic flow sheet that target image of the present invention asks method second embodiment in the center of circle.Ask method first embodiment in the center of circle based on target image of the present invention, in the present embodiment, above-mentioned steps S20 comprises:
Step S201, in described original image each pixel eight neighborhood in, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
With reference to the exemplary plot that Fig. 4, Fig. 4 are the pixel eight neighborhood of image.Generally, the edge of image with the difference that all can exist between background image on gray-scale value, as long as find pixel corresponding to this species diversity substantially can determine the border of an image roughly and background image, the also marginal point of i.e. image.In the present embodiment, also be in the absolute value of gray-scale value difference of upper and lower two pixels, left and right two pixel, bottom right, upper left two pixel and lower-left upper right two pixel, as long as have, namely any one absolute value is non-vanishing illustrates the difference that there is gray-scale value, thus tentatively can judge the marginal point some gradient directions that this pixel is corresponding existing image.
Otherwise namely the if there is no difference of pixel eight neighborhood gray-scale value, also illustrate that this pixel is same gray-scale value with the pixel of its surrounding neighbors, also just say and belong to an image, now need the difference of the eight neighborhood gray-scale value continuing to search next pixel again.
In the present embodiment, for the calculating of the eight neighborhood gray scale difference value of pixel, for improving the precision of calculating further, the present invention calculates experience according to reality, need to carry out small size correction to the calculating of the oblique two pairs of corresponding neighborhood gray-scale values of pixel, such as, when calculating the absolute value of gray-scale value difference of bottom right, upper left two pixel and lower-left upper right two pixel, need all to be multiplied by preferred modified value 0.7, but being not limited to modified value is 0.7, specifically can set according to inventing the applicable situation such as technical field, computing method.
Step S202, carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
According to the measuring and calculating of above-mentioned steps S201, when scanning and calculating difference on gray-scale value of each neighborhood of pixels on view picture original image, identify for there is the pixel of difference on gray-scale value, such as, identify with "-1 ".Can determine the maximum magnitude of target image described in described original image by searching the pixel being designated "-1 ", be also the edge of target image.
In the present embodiment, by the computing method of pixel eight neighborhood gray-scale value difference, achieve the object separated from original image by target image fast, simultaneously, also improve the computing method of original eight neighborhood gray-scale value difference in this enforcement, carried out revising also improving the precision determined at target image edge while ensureing computing velocity to computation process.For convenience of quick position, in the present embodiment, the pixel that be there are differences by gray-scale value is labeled as "-1 ", to determine the maximum magnitude of target image described in described original image, is also the edge of target image.
Further, with reference to Fig. 5-Fig. 6, Fig. 5 is the schematic flow sheet that target image of the present invention asks method the 3rd embodiment in the center of circle; Fig. 6 is the exemplary plot of the round spot edge that the present invention extracts.Ask method first embodiment in the center of circle based on target image of the present invention, in the present embodiment, above-mentioned steps S30 comprises:
Step S301, calculates the mean flow rate of described original image;
The brightness of image specifically refers to the intensity of image pixel, and its scope is 0 ~ 255, and wherein black is the darkest, and white is the brightest, and black represents with 0, and white represents with 255.Each pixel has corresponding brightness, and the calculating of brightness of image has many algorithms, can be calculated by the gray-scale value calculating pixel.Because the R (red of gray level image, red), G (green, green) and B (blue, blue) three components are all equal, therefore the mean value can getting three components and value carrys out the mean flow rate of computed image, but these just more rough computing method, also can increase correction weights and calculate, such as, gray-scale value Gray=0.11*R+0.59*G+0.3*B.
Step S302, in the maximum magnitude in described target image region, according to the mean flow rate of described original image, determines the edge of each described round spot image.
By the mean flow rate of image, image can be divided into the pixel group being greater than mean flow rate and pixel group two parts being less than or equal to mean flow rate, thus can to the binary conversion treatment of image, such as the pixel value of the pixel group being greater than mean flow rate is set as white (or black), the pixel value being less than or equal to the pixel group of mean flow rate is set as black (or white).By the binary conversion treatment of image, thus can by separation of images out.Then using mean flow rate as analytical standard, for the pixel of the gray-scale value below mean flow rate, each round spot peak, minimum point, Far Left, the standard of rightmost around the square frame scope formed as screening circle spot center of circle starting point in target image region.
For promoting the computational accuracy of round spot edge further, employ the Computation schema of threshold value iteration in the present embodiment: the gray feature based on image calculates one or more gray threshold, and by the gray-scale value of pixel each in image compared with threshold value, finally pixel is assigned in suitable classification according to comparative result.
In the present embodiment, the circular step of threshold value iteration is as follows:
Step 1: the mean flow rate calculating view picture original image, for global threshold selects an initial mean flow rate estimated value T
1, described mean flow rate is the segmentation gray-scale value of image;
Computing formula is:
Wherein, iBrightRate is the specific brightness of image, gets the dark rate that preset value 33, iDarkRate is image, gets preset value 67; T
minfor the last time is split lower than the mean flow rate in the pixel group of mean flow rate in image, T
maxfor the last time is split in image higher than the mean flow rate in the pixel group of mean flow rate.Wherein, when first time is split, T
minget the minimum gradation value of original image, T
maxget the maximum gradation value of original image.
Step 2: use T
1carry out first time Iamge Segmentation, thus produce the mean flow rate T that two groups of pixels also calculate two groups of pixels respectively
minand T
max.
Step 3: by T in step 2
minand T
maxsubstitute in the computing formula in step 1 to generate new segmentation threshold T
n, wherein n be greater than 1 positive integer, be preferably taken as n=100 in the present embodiment, the mean flow rate that namely iteration obtains for 100 times is optimum segmentation threshold.
In above iterative process, the crucial part calculated is to select which type of threshold value improvement strategy, and the improvement of good threshold value is measured possess two features, and one is can Fast Convergent, two is in each iterative process, and the new threshold value produced is better than last threshold value.In the present embodiment using the specific brightness (33) of image and dark rate (67) as calculating weights, be better than the computational accuracy of all value 0.5 of specific brightness and dark rate in existing iterative computation.
In the present embodiment, adopt and ask the mode of mean picture brightness to ask the edge of each round spot in target image, by mean flow rate to be calculated the optimum segmentation threshold value of round spot as the mode that threshold value carries out iteration, thus determine the edge of each round spot image.Said method not only can compress the data volume of calculating greatly, and also greatly simplify treatment and analysis step, improves the speed and precision that calculate.
Further, ask the schematic flow sheet of method the 3rd embodiment in the center of circle based on the invention described above target image, in the present embodiment, adopt the mean flow rate of histogrammic method computed image.
What histogram described is the intensity profile curve of image in picture indication range.Histogram refers to the means of the exposure degree of accuracy that to Show Picture with graphical parameter, and what it described is the intensity profile curve of image in picture indication range.The display of the histogram left side be the shadow information of image, the medium tone information of middle display image, the right then shows the highlight information of image.Histogrammic transverse axis from left to right represents photo from black (dark portion) to the pixel quantity of white (highlights), and horizontal axis is 256 grades of gray scale coordinates, and the value of most dark place, the left side is 0, and the value of the right high light is 255.Histogrammic vertical axis represents the quantity of pixel under set-point, and Pixel Information is more at this value that is more just to represent picture.
In Digital Image Processing, histogram is the most simply and the most useful instrument, can say, to the analysis of image and observation until form an effective disposal route, all be unable to do without histogram.Histogram is the function of gray level, description be the frequency that the number of pixels of this gray level in image or this gray-level pixels occur.In the present embodiment, adopt the mean flow rate of histogrammic mode computed image to simplify computation process, improve arithmetic speed.
Further, reference Fig. 7, Fig. 7 is the high-level schematic functional block diagram that target image of the present invention asks device first embodiment in the center of circle.In the present embodiment, described target image asks the device in the center of circle to comprise:
Pretreatment module 10, for reading original image and carrying out gaussian filtering process to described original image, described original image comprises target image region;
In the coordinate system of image acquiescence, bitmap file captured when pretreatment module 10 first reads camera calibration, to ensure that the bitmap graphics created by certain application program can be loaded by other application program or show.For accelerating the processing speed of pre-set image, gray processing process can be carried out according to the form of bitmap pixels and obtaining gray level image.Then, pretreatment module 10 adopts gaussian filtering method to the smoothing process of pre-set image again to improve the image quality of image and to obtain pending image.
Target image edge determination module 20, for carrying out edge extracting to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
Described original image is again after gaussian filtering process, and target image edge determination module 20 carries out edge extracting to be separated from the background of described original image by scaling board image.Have a lot to the method for Image Segmentation Using, the present invention is under the prerequisite considering computing velocity and degree of accuracy, and in conjunction with follow-up calculation procedure and method, in this step, preferably by rim detection, the method for image border is extracted for the dividing method of scaling board image.
Circle spot image border determination module 30, in the maximum magnitude in described target image region, determines the edge of each described round spot image;
According to the result of calculation of target image edge determination module 20, determine the maximum magnitude of target image, in the maximum magnitude of target image, circle spot image border determination module 30 continues the edge determining each described round spot image.Determine that the method at the edge of circle spot image has a lot, such as comparatively common Roberts Cross operator, Prewitt operator, Sobel operator, Canny operator, Kirsch operator etc., often kind of method has its merits and demerits, when carrying out justifying spot identification, select compatibly round spot center identification method extremely important, it directly affects the precision of measurement.
Central coordinate of circle calculation check module 40, for the coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval.
The coordinate corresponding according to the edge justifying each round spot image that spot image border determination module 30 is determined, in the present embodiment, central coordinate of circle calculation check module 40 calculates the central coordinate of circle of same round spot image respectively by ellipse fitting method and gravity model appoach, and then whether difference between the coordinate in described two centers of circle of corresponding verification is respectively within tolerance interval.Such as, the central coordinate of circle that gravity model appoach calculates is (X
1, Y
1)=(3.1,5.7), the central coordinate of circle that ellipse fitting method calculates is (X
2, Y
2)=(6.5,9.7), then coordinate 1 (X
1, Y
1) and coordinate 2 (X
2, Y
2) between the absolute value of difference be respectively | X
1-X
2|=3.4, | Y
1-Y
2|=4.For ensureing the accuracy of result of calculation further, the present invention adopts the center of circle of two kinds of methods to circle spot to solve to limit the difference that both obtain the center of circle, if the difference difference of the two is too large, then this center of circle is not the correct center of circle, if within the acceptable range, then illustrate that this center of circle is the center of circle of needs.
Module 50 is preserved in the center of circle, for when the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserves described central coordinate of circle.
In the present embodiment preferably 0.6 pixel as the standard of the absolute value of difference corresponding to measurement two kinds of methods.Such as, as | X
1-X
2| <=0.6 and | Y
1-Y
2| during <=0.6, then the result accepting to draw is correct central coordinate of circle and preserves.
In the present embodiment, pretreatment module 10 first reads original image and gaussian filtering pre-service, and then target image is separated to reduce the scope of searching calculating further by target image edge determination module 20 again from pre-set image; Secondly, circle spot image border determination module 30 determines the edge of each circle spot again from target image; The set of coordinates that last central coordinate of circle calculation check module 40 is corresponding according to the round spot edge determined again, calculates the center of circle of this circle spot respectively by gravity model appoach and ellipse fitting method.For promoting speed and the precision of measuring and calculating further, in conjunction with computing velocity and accuracy requirement in the present embodiment, the computing method adapted are selected in the different measuring and calculating stages, in the hope of reaching the balance in computing velocity and precision, thus provide a kind of and more calculate fast and meet the combination computing method asking the round spot center of circle of accuracy requirement.
In addition, for ensureing the accuracy of result of calculation further, the present invention obtains the difference in the center of circle by asking the central coordinate of circle of round spot to two kinds of methods with both restrictions, if the difference difference of the two is too large, then illustrate that this center of circle is not the correct center of circle, and then calculate the central coordinate of circle of next circle spot; If within the acceptable range, then illustrate that this center of circle is the center of circle of needs, preserve the coordinate in the described center of circle.The image in different application field asks the center of circle, its requirement for computing method is all not quite similar, simultaneously, different computing method ask its effect calculated of the center of circle also all different for the image in different technologies field, but various method is mainly weighed from the speed calculated and precision two aspects for the requirement of the result calculated.In the present embodiment, according to the feature of place of the present invention application, and combine the speed of calculating and considering of precision two aspect, present invention improves over the method in original computed image center of circle and define a set of method being applicable to comprise the technical applications computed image center of circle that automobile 3D four-wheel detects.
Further, reference Fig. 8, Fig. 8 is the high-level schematic functional block diagram that target image of the present invention asks device second embodiment in the center of circle.Ask device first embodiment in the center of circle based on target image of the present invention, in the present embodiment, target image edge determination module 20 comprises:
Neighborhood gray-scale value difference computational unit 201, in eight neighborhood for each pixel in described original image, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
Generally, the edge of image with the difference that all can exist between background image on gray-scale value, as long as find pixel corresponding to this species diversity substantially can determine the border of an image roughly and background image, the also marginal point of i.e. image.In the present embodiment, also be in the absolute value of gray-scale value difference of upper and lower two pixels, left and right two pixel, bottom right, upper left two pixel and lower-left upper right two pixel, as long as have, namely any one absolute value is non-vanishing illustrates the difference that there is gray-scale value, thus tentatively can judge the marginal point some gradient directions that this pixel is corresponding existing image.
Otherwise namely the if there is no difference of pixel eight neighborhood gray-scale value, also illustrate that this pixel is same gray-scale value with the pixel of its surrounding neighbors, also just say and belong to an image, now need the difference of the eight neighborhood gray-scale value continuing to search next pixel again.
In the present embodiment, for the calculating of the eight neighborhood gray scale difference value of pixel, for improving the precision of calculating further, the present invention calculates experience according to reality, need to carry out small size correction to the calculating of the oblique two pairs of corresponding neighborhood gray-scale values of pixel, such as, when calculating the absolute value of gray-scale value difference of bottom right, upper left two pixel and lower-left upper right two pixel, need all to be multiplied by preferred modified value 0.7, but being not limited to modified value is 0.7, specifically can set according to inventing the applicable situation such as technical field, computing method.
Screening identify unit 202, for carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
When scanning and calculating difference on gray-scale value of each neighborhood of pixels on view picture original image, screening identify unit 202 identifies there is the pixel of difference on gray-scale value, such as, identify with "-1 ".Can determine the maximum magnitude of target image described in described original image by searching the pixel being designated "-1 ", be also the edge of target image.
In the present embodiment, by the computing method of pixel eight neighborhood gray-scale value difference, achieve the object separated from original image by target image fast, simultaneously, also improve the computing method of original eight neighborhood gray-scale value difference in this enforcement, carried out revising also improving the precision determined at target image edge while ensureing computing velocity to computation process.For convenience of quick position, in the present embodiment, the pixel that be there are differences by gray-scale value is labeled as "-1 ", to determine the maximum magnitude of target image described in described original image, is also the edge of target image.
Further, reference Fig. 9, Fig. 9 is the high-level schematic functional block diagram that target image of the present invention asks device the 3rd embodiment in the center of circle.Ask device first embodiment in the center of circle based on target image of the present invention, in the present embodiment, circle spot image border determination module 30 comprises:
Average luminance computing unit 301, for the mean flow rate of computed image;
The brightness of image specifically refers to the intensity of image pixel, and its scope is 0 ~ 255, and wherein black is the darkest, and white is the brightest, and black represents with 0, and white represents with 255.Each pixel has corresponding brightness, and the calculating of brightness of image has many algorithms, can be calculated by the gray-scale value calculating pixel.Because the R (red of gray level image, red), G (green, green) and B (blue, blue) three components are all equal, therefore the mean value can getting three components and value carrys out the mean flow rate of computed image, but these just more rough computing method, also can increase correction weights and calculate, such as, gray-scale value Gray=0.11*R+0.59*G+0.3*B.
Circle spot image border determining unit 302, in the maximum magnitude in described target image region, according to the mean flow rate of described original image, determines the edge of each described round spot image.
By the mean flow rate of image, image can be divided into the pixel group being greater than mean flow rate and pixel group two parts being less than or equal to mean flow rate, thus can to the binary conversion treatment of image, such as the pixel value of the pixel group being greater than mean flow rate is set as white (or black), the pixel value being less than or equal to the pixel group of mean flow rate is set as black (or white).By the binary conversion treatment of image, thus can by separation of images out.Then using mean flow rate as analytical standard, for the pixel of the gray-scale value below mean flow rate, each round spot peak, minimum point, Far Left, the standard of rightmost around the square frame scope formed as screening circle spot center of circle starting point in target image region.
For promoting the computational accuracy of round spot edge further, employ the Computation schema of threshold value iteration in the present embodiment: the gray feature based on image calculates one or more gray threshold, and by the gray-scale value of pixel each in image compared with threshold value, finally pixel is assigned in suitable classification according to comparative result.
In the present embodiment, the circular step of threshold value iteration is as follows:
Step 1: the mean flow rate calculating view picture original image, for global threshold selects an initial mean flow rate estimated value T
1, described mean flow rate is the segmentation gray-scale value of image;
Computing formula is:
Wherein, iBrightRate is the specific brightness of image, gets the dark rate that preset value 33, iDarkRate is image, gets preset value 67; T
minfor the last time is split lower than the mean flow rate in the pixel group of mean flow rate in image, T
maxfor the last time is split in image higher than the mean flow rate in the pixel group of mean flow rate.Wherein, when first time is split, T
minget the minimum gradation value of original image, T
maxget the maximum gradation value of original image.
Step 2: use T
1carry out first time Iamge Segmentation, thus produce the mean flow rate T that two groups of pixels also calculate two groups of pixels respectively
minand T
max.
Step 3: by T in step 2
minand T
maxsubstitute in the computing formula in step 1 to generate new segmentation threshold T
n, wherein n be greater than 1 positive integer, be preferably taken as n=100 in the present embodiment, the mean flow rate that namely iteration obtains for 100 times is optimum segmentation threshold.
In above iterative process, the crucial part calculated is to select which type of threshold value improvement strategy, and the improvement of good threshold value is measured possess two features, and one is can Fast Convergent, two is in each iterative process, and the new threshold value produced is better than last threshold value.In the present embodiment using the specific brightness (33) of image and dark rate (67) as calculating weights, be better than the computational accuracy of all value 0.5 of specific brightness and dark rate in existing iterative computation.
In the present embodiment, adopt and ask the mode of mean picture brightness to ask the edge of each round spot in target image, by mean flow rate to be calculated the optimum segmentation threshold value of round spot as the mode that threshold value carries out iteration, thus determine the edge of each round spot image.Said method not only can compress the data volume of calculating greatly, and also greatly simplify treatment and analysis step, improves the speed and precision that calculate.
Further, ask device the 3rd embodiment in the center of circle based on the invention described above target image, in the present embodiment, target image of the present invention asks the device in the center of circle to adopt the mean flow rate of histogrammic method computed image.
What histogram described is the intensity profile curve of image in picture indication range.Histogram refers to the means of the exposure degree of accuracy that to Show Picture with graphical parameter, and what it described is the intensity profile curve of image in picture indication range.The display of the histogram left side be the shadow information of image, the medium tone information of middle display image, the right then shows the highlight information of image.Histogrammic transverse axis from left to right represents photo from black (dark portion) to the pixel quantity of white (highlights), and horizontal axis is 256 grades of gray scale coordinates, and the value of most dark place, the left side is 0, and the value of the right high light is 255.Histogrammic vertical axis represents the quantity of pixel under set-point, and Pixel Information is more at this value that is more just to represent picture.
In Digital Image Processing, histogram is the most simply and the most useful instrument, can say, to the analysis of image and observation until form an effective disposal route, all be unable to do without histogram.Histogram is the function of gray level, description be the frequency that the number of pixels of this gray level in image or this gray-level pixels occur.In the present embodiment, adopt the mean flow rate of histogrammic mode computed image to simplify computation process, improve arithmetic speed.
These are only the preferred embodiments of the present invention; not thereby the scope of the claims of the present invention is limited; every utilize instructions of the present invention and accompanying drawing content to do equivalent structure or equivalent flow process conversion; or be directly or indirectly used in other relevant technical fields, be all in like manner included in scope of patent protection of the present invention.
Claims (6)
1. target image asks the method in the center of circle, and described target image comprises at least one circle spot image-region, and it is characterized in that, described target image asks the method in the center of circle to comprise:
Read original image and carry out gaussian filtering process to described original image, described original image comprises target image region;
Edge extracting is carried out to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
In the maximum magnitude in described target image region, determine the edge of each described round spot image;
The coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval;
When the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserve described central coordinate of circle.
2. target image according to claim 1 asks the method in the center of circle, it is characterized in that, describedly carries out edge extracting to the described original image after described gaussian filtering process, determines that the maximum magnitude in target image region described in described original image comprises:
In described original image each pixel eight neighborhood in, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
Carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
3. target image as claimed in claim 1 or 2 asks the method in the center of circle, it is characterized in that, described in the maximum magnitude in described target image region, determines that the edge of each described round spot image comprises:
Calculate the mean flow rate of described original image;
In the maximum magnitude in described target image region, according to the mean flow rate of described original image, determine the edge of each described round spot image.
4. target image asks the device in the center of circle, and described target image comprises at least one circle spot image, and it is characterized in that, described target image asks the device in the center of circle to comprise:
Pretreatment module, for reading original image and carrying out gaussian filtering process to described original image, described original image comprises target image region;
Target image edge determination module, for carrying out edge extracting to the described original image after described gaussian filtering process, determines the maximum magnitude in target image region described in described original image;
Circle spot image border determination module, in the maximum magnitude in described target image region, determines the edge of each described round spot image;
Central coordinate of circle calculation check module, for the coordinate corresponding according to the edge of described round spot image, calculate the central coordinate of circle of same described round spot image respectively by ellipse fitting method and gravity model appoach and the absolute value verifying the difference between described two central coordinate of circle whether within tolerance interval;
Central coordinate of circle preserves module, for when the absolute value of the difference between described two central coordinate of circle is within tolerance interval, preserves described central coordinate of circle.
5. target image according to claim 4 asks the device in the center of circle, it is characterized in that, described target image edge determination module comprises:
Neighborhood gray-scale value difference computational unit, in eight neighborhood for each pixel in described original image, calculate respectively the absolute value of gray-scale value difference of up and down two pixels corresponding with this pixel, the absolute value of the gray-scale value difference of left and right two pixel, the absolute value of the gray-scale value difference of bottom right, upper left two pixel and the gray-scale value difference of lower-left upper right two pixel absolute value and ask each absolute value and value;
Screening identify unit, for carrying out described absolute value corresponding for pixel each in described original image screening with value and identifying coordinate corresponding to the pixel non-vanishing with value of described absolute value, wherein, the coordinate that the pixel non-vanishing with value of described absolute value is corresponding around scope be the maximum magnitude in described target image region.
6. the target image as described in claim 4 or 5 asks the device in the center of circle, it is characterized in that, described round spot image border determination module comprises:
Average luminance computing unit, for calculating the mean flow rate of described original image;
Circle spot image border determining unit, in the maximum magnitude in described target image region, according to the mean flow rate of described original image, determines the edge of each described round spot image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510005605.4A CN104574312A (en) | 2015-01-06 | 2015-01-06 | Method and device of calculating center of circle for target image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510005605.4A CN104574312A (en) | 2015-01-06 | 2015-01-06 | Method and device of calculating center of circle for target image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104574312A true CN104574312A (en) | 2015-04-29 |
Family
ID=53090293
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510005605.4A Pending CN104574312A (en) | 2015-01-06 | 2015-01-06 | Method and device of calculating center of circle for target image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104574312A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109872365A (en) * | 2019-02-20 | 2019-06-11 | 上海鼎盛汽车检测设备有限公司 | 3D four-wheel position finder destination disk image-recognizing method |
| CN109978773A (en) * | 2017-12-27 | 2019-07-05 | 浙江宇视科技有限公司 | Image processing method and device |
| CN110332894A (en) * | 2019-07-10 | 2019-10-15 | 中国地质大学（武汉） | A kind of untouchable measurement method of dam surface displacement based on binocular vision |
| CN110660107A (en) * | 2019-08-23 | 2020-01-07 | 贝壳技术有限公司 | Plane calibration plate, calibration data acquisition method and system |
| CN112233076A (en) * | 2020-09-30 | 2021-01-15 | 石家庄铁道大学 | Structural vibration displacement measurement method and device based on red round target image processing |
| CN112581374A (en) * | 2019-09-29 | 2021-03-30 | 深圳市光鉴科技有限公司 | Speckle sub-pixel center extraction method, system, device and medium |
| CN113538479A (en) * | 2020-04-20 | 2021-10-22 | 深圳市汉森软件有限公司 | Image edge processing method, device, equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009259036A (en) * | 2008-04-17 | 2009-11-05 | Sharp Corp | Image processing device, image processing method, image processing program, recording medium, and image processing system |
| CN101650828A (en) * | 2009-09-07 | 2010-02-17 | 东南大学 | Method for reducing random error of round object location in camera calibration |
-
2015
- 2015-01-06 CN CN201510005605.4A patent/CN104574312A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009259036A (en) * | 2008-04-17 | 2009-11-05 | Sharp Corp | Image processing device, image processing method, image processing program, recording medium, and image processing system |
| CN101650828A (en) * | 2009-09-07 | 2010-02-17 | 东南大学 | Method for reducing random error of round object location in camera calibration |
Non-Patent Citations (7)
| Title |
|---|
| 冉冉飘飘: "基于直接线性变换的单目定标模型", 《HTTP://WENKU.BAIDU.COM/VIEW/DA5A2C35EEFDC8D376EE327A.HTML》 * |
| 尚涛 等: "《古代建筑保护数字化技术》", 31 August 2009, 湖北科学技术出版社 * |
| 杨爽: "基于圆形特征的摄像机标定方法的研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 王术彬 等: "基于几何特征的圆心定位方法研究与应用", 《电子设计工程》 * |
| 童静: "基于数字图像的激光光斑的提取与位置检测方法", 《中国优秀硕士学文论文全文数据库 工程科技II辑》 * |
| 赵荣椿 等: "《数字图像处理与分析》", 30 April 2013, 清华大学出版社 * |
| 黄富瑜 等: "全向激光探测系统中光斑精确定位方法研究", 《激光与红外》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978773A (en) * | 2017-12-27 | 2019-07-05 | 浙江宇视科技有限公司 | Image processing method and device |
| CN109872365A (en) * | 2019-02-20 | 2019-06-11 | 上海鼎盛汽车检测设备有限公司 | 3D four-wheel position finder destination disk image-recognizing method |
| CN110332894A (en) * | 2019-07-10 | 2019-10-15 | 中国地质大学（武汉） | A kind of untouchable measurement method of dam surface displacement based on binocular vision |
| CN110660107A (en) * | 2019-08-23 | 2020-01-07 | 贝壳技术有限公司 | Plane calibration plate, calibration data acquisition method and system |
| CN112581374A (en) * | 2019-09-29 | 2021-03-30 | 深圳市光鉴科技有限公司 | Speckle sub-pixel center extraction method, system, device and medium |
| CN113538479A (en) * | 2020-04-20 | 2021-10-22 | 深圳市汉森软件有限公司 | Image edge processing method, device, equipment and storage medium |
| CN113538479B (en) * | 2020-04-20 | 2023-07-14 | 深圳市汉森软件有限公司 | Image edge processing method, device, equipment and storage medium |
| CN112233076A (en) * | 2020-09-30 | 2021-01-15 | 石家庄铁道大学 | Structural vibration displacement measurement method and device based on red round target image processing |
| CN112233076B (en) * | 2020-09-30 | 2022-12-13 | 石家庄铁道大学 | Structural vibration displacement measurement method and device based on red round target image processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104574312A (en) | Method and device of calculating center of circle for target image | |
| EP2085928B1 (en) | Detection of blobs in images | |
| EP2811423A1 (en) | Method and apparatus for detecting target | |
| CN110210448B (en) | Intelligent face skin aging degree identification and evaluation method | |
| US11676301B2 (en) | System and method for efficiently scoring probes in an image with a vision system | |
| CN109034017A (en) | Head pose estimation method and machine readable storage medium | |
| CN109816673A (en) | A kind of non-maxima suppression, dynamic threshold calculates and method for detecting image edge | |
| CN104966285B (en) | A kind of detection method of salient region | |
| EP3358526A1 (en) | System and method for scoring color candidate poses against a color image in a vision system | |
| CN106407924A (en) | Binocular road identifying and detecting method based on pavement characteristics | |
| CN109815822B (en) | Patrol diagram part target identification method based on generalized Hough transformation | |
| CN105718931B (en) | System and method for determining clutter in acquired images | |
| CN107203981A (en) | A kind of image defogging method based on fog concentration feature | |
| CN102236894A (en) | Circle detection method and device | |
| CN110189347B (en) | Method and terminal for measuring volume of object | |
| CN103473776A (en) | Method and system for comparing image definition and automatic focusing control method | |
| CN112991374B (en) | Canny algorithm-based edge enhancement method, canny algorithm-based edge enhancement device, canny algorithm-based edge enhancement equipment and storage medium | |
| CN111354047B (en) | Computer vision-based camera module positioning method and system | |
| CN111222507A (en) | Automatic identification method of digital meter reading and computer readable storage medium | |
| CN114299137A (en) | Laser spot center positioning method and test system | |
| CN106600615A (en) | Image edge detection algorithm evaluation system and method | |
| CN101930600A (en) | Composite second-order fractional order signal processing-based edge detection method | |
| CN116125489A (en) | Indoor object three-dimensional detection method, computer equipment and storage medium | |
| US20210390286A1 (en) | Measuring Quality of Depth Images in Real Time | |
| CN111161251B (en) | Method and device for calculating definition of face image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150429 |