CN1126649C - Robot's visual sensor based on small-wave transform optics - Google Patents

Abstract

A robot vision sensor based on wavelet transform optics is characterized in that an imaging objective lens, a liquid crystal light valve, a polarization beam splitter prism, a collimator lens, a beam expander lens, and a semiconductor laser are fixed from bottom to top in the vertical tube of the housing , the liquid crystal light valve is at the focal length of the imaging objective lens. Two Fourier lenses, an electrical addressing spatial light modulator, and a CCD optoelectronic coupling device are fixed inside the inclined tube of the housing from the inside to the outside. The electrical addressing spatial light modulator is installed in front of the two Fourier lenses, back focal plane. The invention installed on the robot can identify the target and realize automatic tracking, has fast information processing speed, good filtering performance and high repetition accuracy.

Description

Realization of robot vision sensor based on wavelet transform optics

The invention belongs to a robot vision sensor, in particular to a robot vision sensor based on wavelet transform optics.

As a reliable signal processing method, wavelet transform has many advantages compared with traditional transform. Good filtering performance, eliminating redundant noise and interference, improving signal-to-noise ratio, no truncation error, high repeatability and good digital data sequence reconstruction characteristics, etc. At present, most of the existing robot vision image processing is realized based on computer parallel processing software, which usually requires expensive workstations. Especially with the increase of the amount of calculation and processing information, the calculation speed is difficult to meet the real-time requirements in actual engineering.

The present invention realizes robot visual sensor based on wavelet transform optics, can combine optical information processing system and computer, establish various wavelet base filtering function libraries suitable for visual image processing, extract the best image features, and obtain the visual sensor The signal is directly read into the computer, and then the robot arm is controlled to realize automatic tracking movement.

The purpose of the present invention is to overcome the deficiencies in the prior art and provide a robot vision sensor based on wavelet transform optics.

Technical solution of the present invention is as follows:

Realizing the structure of the robot vision sensor based on wavelet transform optics is that an imaging objective lens (1) is fixed at the lower port in the vertical tube of the housing (7), and a liquid crystal light valve (1) is fixed at the focal length of the top of the imaging objective lens (1). 2), the top of the liquid crystal light valve (2), the intersection of the casing vertical cylinder and the axis of the inclined cylinder is installed with a polarization beam splitter prism (3), and a collimator lens (4) is fixed above the polarization beam splitter prism (3); A semiconductor laser (6) is fixed at the upper port in the casing (7), and a beam expander lens (5) is fixed between the semiconductor laser (6) and the collimating lens (5). A Fourier lens (8) and a Fourier lens (10) are fixed in the inclined cylinder of the casing (7), and its position should ensure that the readout light of the polarizing beam splitter prism (3) is incident on the Fourier lens (8). Inside, the rear focal plane of the Fourier lens (8) and the front focal plane of the Fourier lens (10) converge in one place, and an electrically addressable spatial light modulator (9) is installed here. A CCD photocoupling device (11) is fixedly installed at the port in the right oblique cylinder of the Fourier lens (10). Imaging objective lens (1), liquid crystal light valve (2), polarization beam splitter prism (3), beam expander lens (4), collimator lens (5), semiconductor laser (6) are on the same axis, polarization beam splitter prism (3 ), the Fourier lens (8), the electrical addressing spatial light modulator (9), the Fourier lens (10) and the CCD photoelectric coupling device (11) are on the same axis. The electrical addressing spatial light modulator (9) and the CCD photoelectric coupling device (11) are respectively connected with the computer (12) by cables.

The resolution of the liquid crystal light valve 2 is 401p/mm, the contrast ratio is 100:1, and the writing time is 40ms.

The splitting angle of the polarizing beam splitting prism is 45-90 degrees.

The power of semiconductor laser 6 is 5-15mW.

The number of pixels of the electrical addressing spatial light modulator 9 is 800×600, the frame frequency is 200Hz, the pixel size is 33×33, the fill factor is 0.27, and the contrast ratio is 200:1.

The model of the CCD photocoupler 11 is DH512-PRO.

The working principle of the robot vision sensor based on wavelet transform optics is as follows: the external object is imaged on the liquid crystal light valve (2) (LCV-Liquid Crystal Light Valve) through the imaging lens (1), that is, on the optical addressable spatial light modulator . The semiconductor laser (6) emits a laser beam with a wavelength of 635 nanometers, which becomes parallel light through a beam expander lens (5) and a collimator lens (4) and shines on the polarization beam splitter prism (3). The polarization splitter prism (3) polarizes and refracts the parallel light onto the liquid crystal light valve (2). The readout light of the liquid crystal light valve (2) returns along the original path, and then enters the first Fourier lens (8) through the polarization beam splitter prism (3). Realize filtering according to the principle of the optical 4f system, that is, the electrical addressing spatial light modulator (9) is placed on the back focal plane of the Fourier lens (8), and the input filter function spectrum is controlled by the computer, and the image reading is completed here. The light spectrum is multiplied. Because the electrical addressing spatial light modulator (9) is on the front focal plane of the Fourier lens (10), the Fourier lens (10) converts the optical function from the frequency domain to the spatial domain, and finally the object image light through wavelet transform The function directly irradiates on the CCD photoelectric coupling device (11). The CCD optoelectronic coupling device (11) communicates with the computer (12), so that through the processing of converting incoherent light into coherent light, the electrical addressing spatial light modulator (9) controlled by the 4f system filtering principle and the computer (12) is used. , realize wavelet transform processing, and its signal is received by CCD photoelectric coupling device (11), and is read into computer (12).

Compared with the prior art, the present invention has the following advantages:

(1) Realize robot vision sensor based on wavelet transform optics Utilize the characteristics of high-speed, parallel and large-capacity optical information processing to improve information processing speed.

(2) Based on wavelet transform optics, the robot vision sensor has good filtering performance, eliminates redundant noise and interference, improves signal-to-noise ratio, has no truncation error, high repeatability and good digital data sequence reconstruction characteristics, etc.

(3) Realization of robot vision based on wavelet transform optics The sensor is a high-tech product of optomechanical integration, and it is also the product of the combination of optics and computer. The computer completes functions such as signal writing, calculation, storage, programming and control, and directly controls the movement of the robot arm. Based on wavelet transform optics, the robot vision sensor is installed on the robot, which can identify the target and realize automatic tracking.

The descriptions of the attached drawings are as follows:

Fig. 1 is a structural schematic diagram of the present invention.

Embodiments of the present invention are described in further detail below in conjunction with the accompanying drawings:

Realize the structure of the robot vision sensor based on wavelet transform optics as shown in Figure 1: the imaging objective lens (1) is fixedly installed by the screw thread at the lower port in the vertical tube of the "T" shaped housing (7), and the imaging objective lens (1) A liquid crystal light valve (2) is fixed with a set screw at the focal length above. A polarizing beamsplitter prism (3) is installed above the liquid crystal light valve (2) and at the intersection of the axis of the "T"-shaped housing vertical cylinder and the inclined cylinder, and the top of the polarizing beamsplitting prism (3) is fixed with a collimator lens (4 ), the semiconductor laser (6) is clamped at the upper port in the housing (7), and a beam expander lens (5) is fixedly installed by threads between the semiconductor laser (6) and the collimating lens (5). The Fourier lens (8) and the Fourier lens (10) are fixed with set screws in the oblique tube of the "T"-shaped housing (7), and its position should ensure that the readout light of the polarization beam splitter (3) is incident into the Fourier lens (8), and the rear focal plane of the Fourier lens (8) merges with the front focal plane of the Fourier lens (10), and an electric addressing device is installed at the intersection of the focal planes The spatial light modulator (9) is fixedly mounted with a CCD photoelectric coupling device (11) by threads at the port in the right oblique barrel of the Fourier lens (10). Imaging objective lens (1), liquid crystal light valve (2), polarization beam splitter prism (3), beam expander lens (4), collimator lens (5), semiconductor laser (6) are on the same axis. The polarization splitter prism (3), the Fourier lens (8), the electrical addressing spatial light modulator (9), the Fourier lens (10) and the CCD photoelectric coupling device (11) are on the same axis. The electrical addressing spatial light modulator (9) and the CCD photoelectric coupling device (11) are respectively connected with the computer (12) by cables.

The resolution of the liquid crystal light valve (2) is 401p/mm, the contrast ratio is 100:1, and the writing time is 40ms.

The beam splitting angle of the polarizing beam splitting prism (3) is 78 degrees.

The power of semiconductor laser (6) is 10mW.

The number of pixels of the electrical addressing spatial light modulator (9) is 800×600, the frame frequency is 200Hz, the pixel size is 33×33, the fill factor is 0.27, and the contrast ratio is 200:1.

The model of CCD photocoupler (11) is DH512-PRO.

When in use, the robot vision sensor based on optical wavelet transform is installed on the flange of the robot end effector, and is connected with the computer system and the robot controller to communicate with each other to control the robot's arm according to the position recognized by the vision sensor. The path is tracked in real time, and the distance between the visual sensor and the measured object is 1 meter.

At the front end of the robot vision sensor based on wavelet transform optics, the interface for installing the zoom lens is made in advance, and after the zoom lens is connected, the vision sensor based on wavelet transform optics of the present invention can acquire long-distance objects and monitor the environment. things to monitor.

Claims (2)

1, a kind of based on wavelet transformation Optical Implementation vision sensor of robot, it is characterized in that: the lower port place is installing imaging object lens (1) in the upright tube of housing (7), the focal length place, top of imaging object lens (1) is installing liquid crystal light valve (2), the top of liquid crystal light valve (2), the intersection of upright tube of housing and oblique tube axis is installed with polarization splitting prism (3), the top of polarization splitting prism (3) is installing collimation lens (4), upper port place in housing (7) is installing semiconductor laser (6), between semiconductor laser (6) and collimation lens (5), installing an extender lens (5), installing fourier lense (8) and fourier lense (10) in the oblique tube of housing (7), its position will guarantee that the light of reading of polarization splitting prism (3) incides in the fourier lense (8), and make the back focal plane of fourier lense (8) and fourier lense (10) front focal plane merge in a place, and electrical addressing spatial light modulator (9) is installed herein, port in the right-hand oblique tube of fourier lense (10) is installing CCD photoelectric coupled device (11), imaging object lens (1), liquid crystal light valve (2), polarization splitting prism (3), extender lens (4), collimation lens (5), semiconductor laser (6) is on same axis, polarization splitting prism (3), fourier lense (8), electrical addressing spatial light modulator (9), fourier lense (10) and CCD photoelectric coupled device (11) are on same axis, and electrical addressing spatial light modulator (9) uses cable to be connected with computer (12) respectively with CCD photoelectric coupled device (11).

2, according to the said vision sensor of robot of claim 1, it is characterized in that: the resolution ratio of liquid crystal light valve (2) is 401p/mm, and contrast is 100: 1, and the write time is 40ms; The light spliting angle of polarization splitting prism (3) is the 45-90 degree; The power of semiconductor laser (6) is 5-15mW; The number of picture elements of electrical addressing spatial light modulator (9) is 800 * 600, and the picture frequency is 200Hz, and picture dot is of a size of 33 * 33, and fill factor, curve factor is 0.27, and contrast is 200: 1.

Images