CN105741296A - Auxiliary calibration method of 360-degre all-visual-angle aerial view panorama travelling crane - Google Patents

Abstract

The invention relates to an auxiliary calibration method of 360-degre all-visual-angle aerial view panorama travelling crane. The auxiliary calibration method comprises the following steps: arranging a checkerboard on the ground, and driving a vehicle to be calibrated to an appointed area on the checkerboard; detecting the coordinate of a gauge point position in the checkerboard of a fisheye distortion image; correcting the coordinate of the fisheye distortion image as the coordinate of a non-fisheye distortion image; and according to the coordinate of the gauge point in the non-fisheye distortion image, adopting a perspective transformation algorithm to obtain an aerial view visual angle coordinate. The auxiliary calibration method is simple in checkerboard layout, is easy in implementation, is simple in a calibration process, realizes automation, does not need artificial participation, is high in positioning precision and is free from personal errors.

Description

Panorama driving auxiliary calibration method is got a bird's eye view at a kind of 360 degree of full visual angles

Technical field

The present invention relates to auxiliary driving field, specifically panorama driving auxiliary calibration method is got a bird's eye view at a kind of 360 degree of full visual angles.

Background technology

Fast development along with image and computer vision technique, increasing technology is applied to vehicle electric field, traditional installs photographic head based on image reverse image system at the tailstock, region limited around the tailstock can only be covered, and the blind area of the short range of vehicle and headstock adds the hidden danger of safe driving undoubtedly, in the narrow urban district blocked up and parking lot, collision and scratch time easily occur.

It is that the mutual coordinated effect by multiple vision sensors then passes through image processing techniques and all image synthesis processed that panorama driving aid system is got a bird's eye view at 360 degree of full visual angles, form the video image at full visual angle around full car, thus avoiding the sight line blind area of vehicle periphery, reduce probability collision occurring and scratching.

Then panorama driving aid system is got a bird's eye view at 360 degree of current full visual angles needs installation personnel manually to debug carrying out image calibration process, calibration process relative complex, elapsed time is longer, and the quality of stitching image is subject to the impact of manual operation, and the requirement of installation personnel is higher.

Summary of the invention

For the deficiencies in the prior art, the present invention provides a kind of calibration process simple, it is not necessary to panorama driving auxiliary calibration method is got a bird's eye view at the artificial 360 degree of full visual angles participated in.

The present invention be the technical scheme is that for achieving the above object

Panorama driving auxiliary calibration method is got a bird's eye view at a kind of 360 degree of full visual angles, comprises the following steps:

Step 1: arrange gridiron pattern on the ground, and vehicle to be calibrated is travelled to gridiron pattern appointment region；

Step 2: labelling point position coordinates in the gridiron pattern of detection flake fault image；

Step 3: be the coordinate without flake fault image by the coordinates correction of flake fault image；

Step 4: according to without the coordinate of labelling point in flake fault image, adopting perspective transform algorithm, obtain getting a bird's eye view angular view coordinate.

Described appointment region is the rectangular area in the middle of described gridiron pattern, and makes vehicle to be calibrated as wherein.

Described gridiron pattern is the rectangular mesh being made up of some row and columns, and in the surrounding specifying region, every side arranges four groups or above black region, and often group black region includes the black rectangle lattice that two points connect.

Described vehicle to be calibrated arranges fish-eye camera in the center of headstock and the tailstock and rearview mirror bottom, left and right.

The Radix Rumicis of described fish-eye camera is 180 degree.

In the gridiron pattern of described detection flake fault image, labelling point position comprises the following steps:

Step 1: picture contrast stretched, strengthens the contrast of gridiron pattern black and white rectangle；

Step 2: calculate the response value of each pixel；

Step 3: slip local extremum detection window, calculates all local extremums and records its response value and coordinate thereof；

Step 4: choose 4 extreme points that response value is maximum, obtains the position coordinates of labelling point in the gridiron pattern of flake fault image.

The response value of each pixel of described calculating, computational methods are: set a rectangular window, rectangular window center is moved to pixel to be calculated, and both horizontally and vertically rectangular window is bisected into quarter, from left to right it is labeled as A, B, be labeled as C, D from top to bottom, then response value computing formula is:

$Num=\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈B}{\Σ}{p}_i\right|+\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈C}{\Σ}{p}_i\right|-\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈D}{\Σ}{p}_i\right|$

Wherein: p_iFor the gray value of pixel, Num is response value.

The width that width is rectangular grid of described local extremum detection window, is highly the height of rectangular grid, slides laterally distance for the wide half of rectangular grid, and longitudinal sliding motion distance is the high half of rectangular grid.

The described coordinates correction by flake fault image is the coordinate process without flake fault image:

$h_A^{\′}=\frac{2r}{\mathrm{\π}}\arctan \left(\frac{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}{2r/\mathrm{\π}}\right)$

$x^{\′}=\left(\frac{h_A^{\′}(x-w/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{w}{2}$

$y^{\′}=\left(\frac{h_A^{\′}(y-h/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{h}{2}$

Wherein, h, w respectively picture altitude and width, r is flake radius, and pixel coordinate in x, y flake fault image respectively, x', y' are respectively without the coordinate of flake fault image；h'_AFor intermediate variable.

The invention have the advantages that and advantage:

1. gridiron pattern layout of the present invention is simple, it is easy to accomplish；

2. calibration process of the present invention simply and realizes automatization, it is not necessary to manually participate in；

3. positioning precision of the present invention is high, it does not have personal error.

Accompanying drawing explanation

Fig. 1 is the method flow diagram of the present invention；

Fig. 2 is the gridiron pattern schematic layout pattern of the present invention；

Fig. 3 is the labelling point position overhaul flow chart of the present invention.

Detailed description of the invention

Below in conjunction with drawings and Examples, the present invention is described in further detail.

If Fig. 1 is the method flow diagram of the present invention.

The critical function that panorama driving aid system is auxiliary driving field is got a bird's eye view at 360 degree of full visual angles, and the present invention comprises the steps of altogether

Step 1: arrange gridiron pattern, and vehicle is travelled to gridiron pattern appointment region；

Step 2: at headstock and center, tailstock portion and left and right rearview mirror 180 degree of fish-eye cameras mounted below；

Step 3: labelling point position in detection gridiron pattern automatically；

Step 4: the distorted image correction that fish-eye camera collects is orthoscopic image；

Step 5: adopt perspective transform algorithm that photographic head is looked squarely view transformation for getting a bird's eye view visual angle；

Step 6: be entire image by four width image mosaic, it is thus achieved that full multi-view image and splicing index map.

Tessellated layout designs is as shown in Figure 2.

Gridiron pattern is made up of the rectangle frames of 31 row, 15 row, and rectangle frame is of a size of 30cm × 30cm, wherein: the 9th walks to 23 row, the 5th row to the parking area that 11 column regions are vehicle to be calibrated；The position of black rectangle frame such as following table:

Sequence number	Line position	Column position	Sequence number	Line position	Column position
						1	2	7	17	15	2
2	2	9	18	15	14
						3	3	6	19	16	3
4	3	10	20	16	13
						5	5	6	21	18	3
6	5	10	22	18	13
						7	6	7	23	19	4
8	6	9	24	19	12
						9	9	2	25	26	7
10	9	14	26	26	9
						11	10	3	27	27	6
12	10	13	28	27	10
						13	12	3	29	29	6
14	12	13	30	29	10
						15	13	4	31	30	7
16	13	12	32	30	9

Table 1

In step 3, reference points detection is as it is shown on figure 3, comprise procedure below:

Picture contrast is stretched, strengthens the contrast of gridiron pattern black and white block；

Calculate the response value of each pixel, circular is: set a sliding window, sliding window center is moved to pixel to be calculated, and both horizontally and vertically sliding window is bisected into quarter, be from left to right labeled as A, B, C, D from top to bottom successively.Response value computing formula is:

$Num=\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈B}{\Σ}{p}_i\right|+\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈C}{\Σ}{p}_i\right|-\left|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈D}{\Σ}{p}_i\right|$

Wherein: p_iGray value for pixel；

What set local extremum detection window is sized to a tessellated size, and it slides laterally distance for the wide half of rectangular grid, and longitudinal sliding motion distance is the high half of rectangular grid；The detection window that slides calculates all local extremums and records its response value and coordinate thereof；

Choose 4 extreme points that response value is maximum, obtain the coordinate of labelling point in gridiron pattern.

In step 4, flake distortion correction, according to the fish-eye camera selected, sets the selected formula of flake correction, and in this patent, employing original image vegetarian refreshments to the mapping equation of correction chart is:

$h_A^{\′}=\frac{2r}{\mathrm{\π}}\arctan \left(\frac{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}{2r/\mathrm{\π}}\right)$

$x^{\′}=\left(\frac{h_A^{\′}(x-w/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{w}{2}$

$y^{\′}=\left(\frac{h_A^{\′}(y-h/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{h}{2}$

Wherein, h, w respectively picture altitude and width, r is flake radius, pixel coordinate in x, y artwork respectively, x', y' respectively correct after the coordinate of image.

In step 5, the angular coordinate obtained is transformed in the orthoscopic image after flake correction, and carry out perspective correction according to these 4 some positions in step 3.

In step 6, splice the image after the 4 width corrections obtained according to the position of photographic head, and record its concordance list, complete calibration process.

Claims (9)

1. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles, it is characterised in that comprise the following steps:

Step 1: arrange gridiron pattern on the ground, and vehicle to be calibrated is travelled to gridiron pattern appointment region；

Step 2: labelling point position coordinates in the gridiron pattern of detection flake fault image；

Step 3: be the coordinate without flake fault image by the coordinates correction of flake fault image；

Step 4: according to without the coordinate of labelling point in flake fault image, adopting perspective transform algorithm, obtain getting a bird's eye view angular view coordinate.

2. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 1, it is characterised in that: described appointment region is the rectangular area in the middle of described gridiron pattern, and makes vehicle to be calibrated as wherein.

3. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 1 and 2, it is characterized in that: described gridiron pattern is the rectangular mesh being made up of some row and columns, in the surrounding specifying region, every side arranges four groups or above black region, and often group black region includes the black rectangle lattice that two points connect.

4. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 1, it is characterised in that: described vehicle to be calibrated arranges fish-eye camera in the center of headstock and the tailstock and rearview mirror bottom, left and right.

5. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 4, it is characterised in that: the Radix Rumicis of described fish-eye camera is 180 degree.

6. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 1, it is characterised in that in the gridiron pattern of described detection flake fault image, labelling point position comprises the following steps:

Step 1: picture contrast stretched, strengthens the contrast of gridiron pattern black and white rectangle；

Step 2: calculate the response value of each pixel；

Step 3: slip local extremum detection window, calculates all local extremums and records its response value and coordinate thereof；

Step 4: choose 4 extreme points that response value is maximum, obtains the position coordinates of labelling point in the gridiron pattern of flake fault image.

7. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 6, it is characterized in that, the response value of each pixel of described calculating, computational methods are: set a rectangular window, rectangular window center is moved to pixel to be calculated, and both horizontally and vertically rectangular window is bisected into quarter, be from left to right labeled as A, B, be labeled as C, D from top to bottom, then response value computing formula is:

$Num=|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈B}{\Σ}{p}_i|+|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈C}{\Σ}{p}_i|-|\underset{i\∈A}{\Σ}{p}_i-\underset{i\∈D}{\Σ}{p}_i|$

Wherein: p_iFor the gray value of pixel, Num is response value.

8. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 6, it is characterized in that, the width that width is rectangular grid of described local extremum detection window, it it is highly the height of rectangular grid, sliding laterally distance for the wide half of rectangular grid, longitudinal sliding motion distance is the high half of rectangular grid.

9. panorama driving auxiliary calibration method is got a bird's eye view at 360 degree of full visual angles according to claim 1, it is characterised in that the described coordinates correction by flake fault image is the coordinate process without flake fault image is:

$h_A^{\′}=\frac{2r}{\mathrm{\π}}arctan\left(\frac{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}{2r/\mathrm{\π}}\right)$

$x^{\′}=\left(\frac{h_A^{\′}(x-w/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{w}{2}$

$y^{\′}=\left(\frac{h_A^{\′}(y-h/2)}{\sqrt{{(x-w/2)}^2+{(y-h/2)}^2}}\right)+\frac{h}{2}$

Wherein, h, w respectively picture altitude and width, r is flake radius, and pixel coordinate in x, y flake fault image respectively, x', y' are respectively without the coordinate of flake fault image；h'_AFor intermediate variable.