CN101324430A - Binocular Ranging Method Based on Similarity Principle - Google Patents

Abstract

The invention belongs to the field of mobile robots and proposes a novel binocular ranging method. The hardware of the ranging method mainly includes two cameras with different focal lengths, an image acquisition card, a camera control motor and a main control computer. According to the similarity principle of triangles, and the relationship formulas of focal length, object distance and image distance, the distance and size calculation formulas of the measured objects are deduced. Two cameras with different focal lengths are used to image the measured object to obtain two images of different scales. The main control computer calculates the different sizes of the measured object in the two images through corresponding algorithms, and substitutes these two different sizes into the calculation The distance and size information of the measured object can be obtained by using the formula. This method enables the robot to have the effect of a telescope and a wide-angle lens at the same time, improving the depth and breadth of its perception of the world. According to the principle formula of the method, the distance information of the object can be calculated without measuring the image distance, thereby saving the use of displacement sensors and reducing the cost.

Description

Binocular Ranging Method Based on Similarity Principle

technical field

The invention belongs to the field of intelligent mobile robots and proposes a novel binocular ranging method.

Background technique

To realize the autonomous movement of intelligent mobile robots, it is inseparable from path planning and real-time obstacle avoidance, and a key technology of real-time obstacle avoidance is to accurately measure the distance from the robot to the obstacle. At present, the commonly used ranging methods are mainly divided into two categories: active ranging and passive ranging.

(1) Active ranging

The active ranging method is to emit energy through a specific device, and the ranging system measures the distance of the object according to the reflected information. It mainly includes the reflected energy method and the ultrasonic time method. The reflected energy method requires a special instrument to emit a beam of light (usually near-infrared light or laser) to the surface of the measured object, and the instrument receives the reflected light energy of the measured object at the same time, and judges the measured object according to the received reflected light energy. Distance: Ultrasonic time method measures the time from the emission of a beam of ultrasound to the reflection back to the instrument to judge the measured distance. The advantage of the active ranging method is that it is less disturbed by the external environment, and its disadvantage is that it requires an additional energy transmitting device, which increases the cost of the equipment, and some may require an energy transmission medium (such as the ultrasonic time method), which is greatly affected by the environment. Its usage is limited.

(2) Passive ranging

Passive ranging method is to measure the distance of the object according to the signal (such as light signal) sent by the measured object itself. It is usually associated with machine vision, mainly including stereo vision ranging method, monocular distance measuring method, passive angle measurement distance method etc. Stereo vision distance measurement is a distance measurement method modeled on humans using binocular perception of distance. The difficulty of this method is to select reasonable matching features and matching criteria to ensure the accuracy of matching, and to make the robot perceive the depth and breadth of the world. Limited; the monocular ranging method introduces a "mask" that meets certain conditions in the optical system, so that the optical transfer function of the imaging system forms a series of zero points that are periodically changed and are related to the distance of the target object. This method requires the target object to have a low Spatial frequency characteristics, at the same time, precision instruments are required to measure the image distance information; the angle measurement passive ranging method is a ranging method proposed by the relevant units of the US Navy, and the ranging is realized by measuring the angle of the target twice. This method requires platform acceleration Cannot be zero.

It can be seen that there are certain shortcomings or deficiencies in the various active ranging and passive ranging methods mentioned above.

The purpose of the present invention is:

Provide a low-cost, high-precision, and easy-to-implement ranging method for intelligent mobile robots, and at the same time improve the depth and breadth of the robot's perception of the world.

Technical scheme of the present invention is:

The binocular ranging method based on the principle of similarity, its hardware mainly includes two cameras with different focal lengths, an image acquisition card, a camera control motor and a main control computer. Two cameras with different focal lengths image a measured object and obtain two images with different scales. The image acquisition card sends the two images to the main control computer, and calculates the distance between the measured object in the two images through the corresponding algorithm. Different sizes. Then, according to the similarity principle of triangles and the relational formula of focal length, object distance and image distance, the distance and size information of the measured object can be calculated.

The advantages of the present invention are:

The new binocular ranging method based on the principle of similarity will use two cameras with different focal lengths in the application, which can enable the robot to have a telescope and a wide-angle lens at the same time, and improve the depth and breadth of the robot's perception of the world. According to the principle formula of the method, the distance information of the object can be calculated without measuring the image distance, thereby saving the use of displacement sensors and reducing the cost. Through a series of experiments, it is also verified that the method has high precision and is easy to implement.

Description of drawings

Figure 1 is a schematic diagram of the new binocular ranging method

Figure 2 is a structural block diagram of the new binocular ranging system

Explanation of symbols in the figure

(1) Number series:

1-lens 1; 2-lens 2; 3-binocular camera; 4-camera steering motor encoder; 5-coder revolutions; 6-master computer; 7-motion planning control system; 8-motion command; - motion motor; 10 - visual information; 11 - image capture card.

(2) Alphabet series:

AB-the measured object; h-the length of the measured object; u-the distance between the measured object and the two lenses; O ₁ -the center of lens one; O ₂ -the center of lens two; f ₁ -the focal length of lens one; f ₂ - focal length of lens 2; A ₁ B ₁ - image of AB through lens 1; A ₂ B ₂ - image of AB through lens 2; v ₁ - image distance of lens 1; v ₂ - image of lens 2 distance; h ₁ - the imaging scale of the measured object through the first lens; h ₂ - the imaging scale of the measured object through the second lens.

Detailed ways

First, let’s introduce the working principle of the new ranging method in detail:

The schematic diagram of the new binocular vision ranging method is shown in Figure 1. In Figure 1, according to the similarity principle of triangles, the following equation is obtained:

$\frac{{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}{{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; h</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}{{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

From formula (1), we can get: $v_1=\frac{h_1{v}_2}{h_2}---\left(2\right)$

From the relationship formula of focal length, object distance and image distance:

$\left\{\begin{array}{c}\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="f"\; filenames="A20081000064400061.png"\; state="\{\{state\}\}">f</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}}\\ \frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="f"\; filenames="A20081000064400061.png"\; state="\{\{state\}\}">f</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; u</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}}\end{array}\right.<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Arranging formulas (2) and (3) to get the calculation formula of object distance:

$u=\frac{f_1{f}_2(h_1-h_2)}{h_1{f}_2-f_1{\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="A20081000064400061.png"\; state="\{\{state\}\}">h</figure-callout>}}_2},$ When f ₁ ≠ f ₂ (4)

And get the calculation formula of the measured object size:

$h=\frac{h_1{h}_2(f_1-f_2)}{h_1{f}_2-f_1{\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="A20081000064400061.png"\; state="\{\{state\}\}">h</figure-callout>}}_2},$ When f ₁ ≠ f ₂ (5)

(1), (2), (3), (4) and (5) where:

h-the length of the measured object AB, corresponding to |AB| in Figure 1; U-the distance from the measured object to the lens, corresponding to |AO ₁ | in Figure 1; h ₁ -the scale of the measured object’s imaging through lens one , corresponding to |A ₁ B ₁ | in Figure 1; h ₂ - the scale of the measured object imaging through lens two, corresponding to |A ₂ B ₂ | in Figure 1; v ₁ - the image distance of lens one imaging, corresponding to Figure 1 |A ₁ O ₁ | in 1; v ₂ - the imaging distance of lens two, corresponding to |B ₂ O ₂ | in Figure 1; f ₁ - the focal length of lens one; f ₂ - the focal length of lens two.

The specific implementation of the ranging method in the application is introduced below:

In practical applications, two cameras with different focal lengths can be used as two lenses. First, the focal lengths f ₁ and f ₂ of the two cameras must be known, and then the imaging dimensions h ₁ and h ₂ of the two cameras can be obtained through the corresponding algorithm. The distance and size information of the measured object can be calculated.

A binocular vision system as shown in Figure 2 can be designed. The two cameras 3 are driven by servo motors, and their viewing angles can reach 360°. The servo motor is equipped with a code disc 4, which is used to calculate the rocking angle and the pitching angle of the camera. The image acquisition card 11 transmits the vision 10 information from the camera to the main control computer 6 for preprocessing, and then integrates the information such as the number of revolutions of the code disc 5, and obtains corresponding instructions after software processing, and the vehicle-mounted motion planning control system 7 sends instructions to The motion motor 9 sends motion commands 8 to control the motion of the robot.

Claims (1)

1. the hardware of this telemetry mainly comprises two camera, image pick-up card, camera control motor and main control computers that focal length is different.According to the leg-of-mutton principle of similitude, and the relation formula of focal length, object distance and image distance can be derived the computing formula of testee distance.

$u=\frac{f_1{f}_2(h_1-h_2)}{h_1{f}_2-f_1{h}_2},$ Work as f ₁≠ f ₂The time

And obtain testee size calculation formula:

$h=\frac{h_1{h}_2(f_1-f_2)}{h_1{f}_2-f_1{h}_2},$ Work as f ₁≠ f ₂The time

Utilize two different cameras of focal length that a testee is carried out imaging, obtain two images that yardstick is different, image pick-up card sends to main control computer with two images, go out the different size of testee in two images by corresponding algorithm computation, in the computing formula with these two different size value substitutions derivations, just can obtain the distance and the dimension information of testee.

Images