CN109141376B - Monocular vision-based wave direction detection method - Google Patents

Abstract

The invention relates to a method for estimating wave direction by utilizing monocular images of waves. Firstly, setting a position point path traversed by an image, selecting straight lines with different slopes at each position point, fitting a straight line according to pixel value data of pixel points superposed in each straight line and the image, and then calculating the size of a pixel value change period by using the pixel value data and the fitted straight lines. And then, calculating the slope of the straight line corresponding to the minimum pixel period obtained by calculation in all the position points to obtain an average value, wherein the average value is the wave slope of the waves in the image. The method provided by the invention can obtain better accuracy in wave direction estimation and has the advantages of small influence of an image shooting angle, and can improve the water surface environment perception capability of a ship when the ship sails in a complex water area.

Description

Monocular vision-based wave direction detection method

Technical Field

The invention belongs to the technical field of water surface environment detection, and particularly relates to a monocular vision wave direction detection method.

Background

When a ship sails in the wind and waves, the sailing resistance encountered is increased. When the ship is sailed by top waves, because the collision period of the surge and the ship is shortened, the collision frequency of the surge is increased, the collision degree is intensified, and the damage to the ship body is increased; when the ship sails perpendicular to the direction of the surge propagation, the generated rolling can cause the life raft, the life boat, the anchor and other decks to be easily bound, loosened and separated by animals, and can fall into the sea when serious; surge resistance can also overload and slow down the marine main engine, can load up electromechanical equipment, and can cause abnormal conditions to occur in such equipment. The above situations all bring difficulties to the ship control and affect the navigation safety.

In order to solve the above problems, the ship needs to know some hydrological information of its own water area in real time to deal with the interference caused by the wind waves, that is, the ship needs to have a certain sensing capability of the water surface environment, and the wave direction detection is an indispensable part of the sensing capability of the water surface environment of the ship. Common wave direction detection methods comprise a threshold segmentation method, a Fourier transform method and a Radon transform method, and the methods have several defects, namely, due to the complex algorithm and large calculation amount, the instantaneity is insufficient; secondly, the method is easily influenced by illumination or shooting angles, and the identification effect is not accurate enough; thirdly, the image to be detected needs high resolution and is high in cost. Therefore, in view of the above disadvantages, there is a need for a wave direction detection method with strong real-time performance and anti-interference of illumination and shooting angle.

Disclosure of Invention

The technical problem to be solved by the present invention is to provide a wave direction detection method based on monocular vision by using a computer image processing technology, according to the phenomenon that in a monocular image, a wave has a strong texture characteristic, the pixel gray value of a wave peak relative to a wave trough is high, any one straight line direction in the image must pass through a plurality of wave peaks and wave troughs, the pixel gray value from the wave peak to the wave trough is periodically changed from high to low in each straight line direction, and the period of the change is shortest in the wave direction.

In order to solve the above technical problems, the present invention provides a wave direction detection method based on monocular vision, which is characterized by comprising the following steps:

step 1: set point of position

Let the size of the image F shot by the monocular camera be M lines, N columns, the ith position point b_iPosition in the image (x)_i,y_i) Comprises the following steps:

wherein i is 1,2, … n; n is N/5, and the total number of the position points is obtained.

Step 2: recording pixel values

Let the current position point be b_iTaking the position point as the center, 36 straight lines with an angle interval of 5 degrees are selected, and the jth line passes through the position point b_i(x_i,y_i) Straight line Y of_jThe equation of (a) is:

Y_j＝tan(5j)(X_j-x_i)+y_i (2)

wherein j is 1,2, … 36; x_j＝1,2,…N；

Straight line Y_jPixel value H of pixel point coinciding with image F_wComprises the following steps:

H_w＝F(X_j,Y_j′) (3)

wherein w is 1,2, … w_j,w_jIs a straight line Y_jThe number of pixels coincident with the image F; y is_j' is Y_jAn integer rounded up; f (x, y) is the pixel value of image F at the (x, y) location.

And step 3: fitting straight line and calculating pixel period

Using least squares method, according to H_wFitting a straight line by the pixel value data in the step (b), finding all the zero-crossing points by an up-crossing zero point method (namely, the point which is intersected with the fitted straight line when the pixel value rises is taken as the up-crossing zero point), and recording the difference value of corresponding abscissas of every two adjacent zero-crossing points as t_pWherein p is 1,2, …, p_j；p_jThe number of the upper zero crossing points is reduced by 1.

All t are compared_pSorting the values in descending order, discarding the value a percent after the sorting value, and leaving t_pIs denoted by t_q(ii) a Wherein q is 1,2, …, q_j；q_j＝p_j(100-a); and a is a fault tolerance ratio coefficient, and a is 1-50, and an empirical value is selected. Then at b_iThe pixel period in the jth straight line direction at the position point is as follows:

wherein i is the ith position point, i is 1,2, …, n; j is the jth line, j being 1,2, …, 36. Step 4: drawing wave-direction lines

Recording the slope of the line with the minimum pixel period in all the lines at the ith position point as K_iThe slope of the wave line

Comprises the following steps:

in the image F taken by the monocular camera, the wave direction line equation representing the wave direction is:

compared with the prior art, the method has higher real-time performance, can quickly acquire the wave direction of the target water area, and is slightly influenced by the shooting angle of the camera; wave direction interference information on an appointed navigation path can be timely and effectively provided for the ship in navigation; the method provided by the invention can obtain better accuracy in wave direction estimation and can improve the water surface environment perception capability of the ship.

The invention also has the advantages of low detection cost and low dependence of the detection device.

Description of the drawings:

fig. 1 is a wave direction detection flow chart based on monocular vision according to the present invention.

The specific implementation mode is as follows:

the following is a preferred embodiment of the present invention and is further described with reference to the accompanying drawings, but the present invention is not limited to the embodiment.

As shown in fig. 1, the monocular vision-based wave direction detection method of the present invention includes the following steps:

step 1: set point of position

The size of the image F shot by monocular vision is M rows, N columns, the ith position point b_iPosition in the image (x)_i,y_i) Comprises the following steps:

wherein i is 1,2, … n; n is N/5, and the total number of the position points is obtained.

Step 2: recording pixel values

Let the current position point be b_iTaking the position point as the center, 36 straight lines with an angle interval of 5 degrees are selected, and the jth line passes through the position point b_i(x_i,y_i) Straight line Y of_jThe equation of (a) is:

Y_j＝tan(5j)(X_j-x_i)+y_i (2)

wherein j is 1,2, … 36; x_j＝1,2,…N；

Straight line Y_jPixel value H of pixel point coinciding with image F_wComprises the following steps:

H_w＝F(X_j,Y_j′) (3)

wherein w is 1,2, … w_j,w_jIs a straight line Y_jThe number of pixels coincident with the image F; y is_j' is Y_jAn integer rounded up; f (x, y) is the pixel value of image F at the (x, y) location.

And step 3: fitting straight line and calculating pixel period

Using least squares method, according to H_wFitting a straight line by the pixel value data in the step (b), finding all the zero-crossing points by an up-crossing zero point method (namely, the point which is intersected with the fitted straight line when the pixel value rises is taken as the up-crossing zero point), and recording the difference value of corresponding abscissas of every two adjacent zero-crossing points as t_pWherein p is 1,2, …, p_j；p_jThe number of the upper zero crossing points is reduced by 1.

All t are compared_pSorting the values in descending order, discarding the value a percent after the sorting value, and leaving t_pIs denoted by t_q(ii) a Wherein q is 1,2, …, q_j；q_j＝p_j(100-a); and a is a fault tolerance ratio coefficient, and a is 1-50, and an empirical value is selected. Then at b_iThe pixel period in the jth straight line direction at the position point is as follows:

wherein i is the ith position point, i is 1,2, …, n; j is the jth line, j is 1,2, …, 36;

and 4, step 4: drawing wave-direction lines

Recording the slope of the line with the minimum pixel period in all the lines at the ith position point as K_iThe slope of the wave line

Comprises the following steps:

in the image F taken by the monocular camera, the wave direction line equation representing the wave direction is:

Claims (1)

1. a wave direction detection method based on monocular vision is characterized by comprising the following steps:

step 1: set point of position

Let the size of the image F shot by the monocular camera be M lines, N columns, the ith position point b_iPosition in the image (x)_i,y_i) Comprises the following steps:

wherein i is 1,2, … n; n is N/5 and is the total number of the position points;

step 2: recording pixel values

Let the current position point be b_iTaking the position point as the center, 36 straight lines with an angle interval of 5 degrees are selected, and the jth line passes through the position point b_i(x_i,y_i) Straight line Y of_jThe equation of (a) is:

Y_j＝tan(5j)(X_j-x_i)+y_i (2)

wherein j is 1,2, … 36; x_j1,2, … N; straight line Y_jPixel value H of pixel point coinciding with image F_wComprises the following steps:

H_w＝F(X_j,Y_j′) (3)

wherein w is 1,2, … w_j,w_jIs a straight line Y_jThe number of pixels coincident with the image F; y is_j' is Y_jAn integer rounded up; f (x, y) is the pixel value of image F at the (x, y) location;

and step 3: fitting straight line and calculating pixel period

Using least squares method, according to H_wFitting a straight line by the pixel value data in the step (2), setting the point crossed with the fitted straight line when the pixel value rises as an ascending zero point, finding all the ascending zero points, and recording the difference value of the corresponding abscissa of every two adjacent ascending zero points as t_pWherein p is 1,2, …, p_j；p_jThe number of the upper span zero points is reduced by 1;

all t are compared_pSorting the values in descending order, discarding the value a percent after the sorting value, and leaving t_pIs denoted by t_q(ii) a Wherein q is 1,2, …, q_j；q_j＝p_j(100-a); a is a fault-tolerant ratio coefficient, a is 1-50, and an empirical value is selected; then at b_iThe pixel period in the jth straight line direction at the position point is as follows:

wherein i is the ith position point, i is 1,2, …, n; j is the jth line, j is 1,2, …, 36;

and 4, step 4: drawing wave-direction lines

Recording the slope of the line with the minimum pixel period in all the lines at the ith position point as K_iThe slope of the wave line

Comprises the following steps:

in the image F taken by the monocular camera, the wave direction line equation representing the wave direction is:

Images