Review of Optical Microvision-Based Precision Positioning Measurement (Invited)

Chenyang Zhao; Jie Xiang; Kai Bian; Zijian Zhu; Qinghong Wan

doi:10.3788/LOP231924

Laser & Optoelectronics Progress, Volume. 61, Issue 2, 0211021(2024)

Review of Optical Microvision-Based Precision Positioning Measurement (Invited)

Chenyang Zhao^*, Jie Xiang, Kai Bian, Zijian Zhu, and Qinghong Wan

School of Mechanical Engineering and Automation, Harbin Institute of Technology (Shenzhen), Shenzhen 518057, Guangdong , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(21)

Fig. 1. Two commonly used illumination models in optical micro-vision-based measurement system. (a) Bright-field imaging model; (b) dark-field imaging model

Download full size

Fig. 2. Dark-field imaging simulation. (a) Simulation of the defect surface and light intensity distribution^[18]; (b) simulation of the etched surface step morphology and near-field light intensity distribution^[21]

Download full size

Fig. 3. Imaging models. (a) Perspective projection model; (b) parallel projection model

Download full size

Fig. 4. Positioning method based on periodic fringe pattern^[35]. (a) Composition of the measurement system; (b) centerline extraction of stripes; (c) processing result of noise filtering; (d) verification of angle measurement

Download full size

Fig. 5. Sampling Moire measurement method. (a) Sampling Moire method principle^[45]; (b) specimen deformation measurement tape^[40]; (c) positioning measurement method using arbitrary periodic array pattern^[46]

Download full size

Fig. 6. Precision positioning methods based on checkerboard pattern. (a) Five-axis platform measurement^[52]; (b) five-axis machine tool calibration based on micro-vision^[42]; (c) six degrees of freedom measurement based on convolutional neural networks^[54]

Download full size

Fig. 7. Positioning measurement methods based on stripe coding pattern. (a) Measurement method based on the Manchester code^[60]; (b) measurement method based on binary phase coding^[61]; (c) measurement method based on M-sequence pseudo random coding^[57]

Download full size

Fig. 8. Two-dimensional coding pattern for visual localization measurement. (a) Pseudo-periodic coding with missing round holes^[68]; (b) interwoven grid-like coding^[59]; (c) missing rectangular coding by André's team^{[58, 62-63]}; (d) missing rectangular encoding by Kim's team^[64]; (e) encoding combined with M-sequence code and checkerboard grid^[67]; (f) shifted circular encoding^[56]

Download full size

Fig. 9. Three typical speckle patterns^[82]. (a) Surface texture; (b) synthetic speckle pattern; (c) interference speckle pattern

Download full size

Fig. 10. Schematic of the fast and robust feature-based positioning (FRFP) method^[92]

Download full size

Fig. 11. Optical microscopic image of the surface microstructure^{[77, 92]}. (a) Experimental result of the polar microstructure surface; (b) partial enlarged view; (c) surface topography measurement of the polar microstructure; (d) simulation result

Download full size

Fig. 12. Biological cell localization technology. (a) Schematic of microdevice for immobilizing mouse embryos and 3 × 3 array of immobilized cells^[96]; (b) micropipette segmentation^[97]

Download full size

Fig. 13. Application fields of micro-vision positioning measurement. (a) Three degrees of freedom nano-positioning^[105]; (b) micro-nano robot control^[106]; (c) wafer alignment^[104]; (d) single-molecule positioning; (e) metallographic detection^[109]; (f) material modification^[110]; (g) drug injection^[108]; (h) sample micromanipulation; (i) cell tracking^[107]

Download full size

Fig. 14. Development route of optical imaging hardware equipment. (a) Natural light source; (b) incandescent lamp; (c) laser light source^[111-112]; (d) nano light source^[113]; (e) spherical lens; (f) Fresnel lens; (g) superlens^[115-116]; (h) photosensitive film; (i) photodiode; (j) CCD and CMOS^[117]; (k) super-resolution detector^[118-119]

Download full size

Fig. 15. Comparison of image matching algorithms. (a) Template matching^[121]; (b) feature point matching^[92]; (c) deep learning matching^[123]

Download full size

Table 1. Different PSF expressions of models

View table

Table 1. Different PSF expressions of models

PSF model	Feature	Expression
Gaussian PSF （3D）	Simple and intuitive	$h (x, y, z) = \frac{1}{σ {(z)}^{2}} e^{- \frac{x^{2} + y^{2}}{2 σ {(z)}^{2}}}$
Defocused PSF （3D）^［26］	Extended depth of field	$h (ω, z) = e^{- σ {(z)}^{2} \cdot ω^{2}} \cdot \frac{s i n ζ (ω, z)}{ζ (ω, z)} \begin{matrix} \cdot \end{matrix} ζ (ω, z) = \frac{z \cdot ω (1 - ω)}{K (z_{i} - z)}$
Born and Wolf PSF （3D）^［27］	Extended depth of field	$h (x, y, z) = {\|C \int_{0}^{1} J_{0} [k \frac{N A}{n} {(x^{2} + y^{2})}^{0.5} ρ] e^{- \frac{1}{2} j k ρ^{2} z {(\frac{N A}{n})}^{2}} ρ d ρ\|}^{2}$
Circular pupil PSF （2D）^［21］	Easy to simulate	$h (x, y) \propto {\|\frac{λ f D}{\sqrt[]{x^{2} + y^{2}}} J_{1} [\frac{2 π}{λ} (\frac{D}{f_{c}} - \frac{2}{\sqrt[]{n^{2} / N A^{2} - 1}}) \sqrt[]{x^{2} + y^{2}}]\|}^{2}$
Square pupil PSF （2D）	Easy to simulate	$h (x, y) \propto {\|\frac{λ f D}{\sqrt[]{x^{2} + y^{2}}} s i n c [\frac{2 π}{λ} (\frac{D}{f_{c}} - \frac{2}{\sqrt[]{n^{2} / N A^{2} - 1}}) \sqrt[]{x^{2} + y^{2}}]\|}^{2}$

Table 2. Optical micro-vision configuration and imaging models
View table
Table 2. Optical micro-vision configuration and imaging models
Optical configuration Projection model
Conventional lens Large depth of field Perspective projection
Small depth of field Parallel projection
Telecentric lens Parallel projection
Confocal Perspective projection

Table 3. Periodic-pattern-based measurement methods and their performance

View table

Table 3. Periodic-pattern-based measurement methods and their performance

Pattern	Method	Preparation	Degrees of freedom	Precision	Range	Resolution	Efficiency
Stripe pattern	Image pyramid^［34］	Quartz	1（x）	12.5 nm	2 μm	‒	‒
	Edge detection^［35］	‒	2（x，θ）	‒	7.46 mrad	0.04 μm，0.29 μrad	‒
	Wavelet transform^［36］	‒	2（x，y）	10 nm	‒	‒	‒
	Delayed Moiré fringes^［37］	Print	2（x， y）	‒	‒	0.036 mm	15 Hz
Array pattern	Monocular 3D reconstruction^［38］	‒	3（α，β，θ）	1"	2π rad	—	‒
	Discrete Fourier analysis^［39］	Etch mask	3（x，y，θ）	<10^-2 pixel	‒	<10 nm	‒
	Discrete Fourier analysis^［18］	Iphone 4s screen	3（x，y，θ）	‒	‒	3.5 nm（x） 8 nm（y）4 μrad（θ）	‒
	Sampling Moiré^［40］	Printed tape	1（x）	<4 μm	‒	‒	‒
Grid pattern	Center point extraction^［41］	Glass grille	3（x，y，θ）	7 μm	220 mm	‒	4 Hz
Checkerboard	Monocular PnP^［42］	‒	4（x，y，z，θ）	<1 μm	<50 μm（x，y） <20 μm（z） 360°（θ）	‒	>10 Hz

Table 4. Pseudo-periodic pattern-based measurement methods and their performance

View table

Table 4. Pseudo-periodic pattern-based measurement methods and their performance

Pattern	Method	Preparation	Degrees of freedom	Precision	Range	Resolution	Efficiency
1D coding pattern	Manchester encoding + periodic grating stripes^［60］	‒	1（x/θ）	‒	100 mm	<1 nm	300 Hz（CPU） 1 MHz（DSP）
	M-sequence pseudo-random coding^［57］	Mask	1（x）	0.79 μm	1310.72 mm	‒	0.3 m/s
	Binary phase encoding^［61］	Quartz mask	1（θ）	0.1″	360°	0.044° （13 bits）	500 Hz
2D coding pattern	Interleaved trellis coding+Fourier analysis^［59］	Photolithography	3（x，y，θ）	18.5 nm	221.33 μm	0.5 nm	30 Hz
	Missing rectangle encoding+Fourier analysis^{［58，62-63］}	Mask	6	1 nm（x，y） 4 μrad（θ）	110 mm（x，y） 10 mm（z）	<1 nm（x，y） < 0.1 mm（z） <1 μrad（θ）， <100 μrad（α，β）	30 Hz（512 pixel×512 pixel）
	Recognition of missing rectangular encoding feature points^［64］	Flat panel display	6	15 μm 0.05°	‒	‒	‒
	Shift cyclic ring coding^{［56，65-66］}	Photolithography	6	1.6 μm（x，y） 1.4 μm（z） 0.0132°	230 mm（x，y）	‒	7 m/min
	M-sequence coding+checkerboard pattern^［67］	Photolithography	2	1 µm	80 mm（x，y）	‒	‒

Table 5. Nonperiodic pattern-based measurement methods and their performance

View table

Table 5. Nonperiodic pattern-based measurement methods and their performance

Pattern	Method	Preparation	Degrees of freedom	Precision	Range	Resolution
Speckle pattern	Spatial gradient^［70］	Computer simulation	2	0.005 pixel	Low	Low
	Newton-Raphson^［71］	Spraying	3	500 μm	Low	Low
	Pro-A^［72］	Surface texture	3	400 nm	Low	High
	CIR^［73］	Surface projection	2	1 nm	Low	High
Microstructurepattern	Template matching^［74］	Ultraprecision machining	2	100 nm	Large	Low
	Pattern recognition^［75］	Ultraprecision machining	2	90 nm	Large	High
	Neural network^［76］	Ultraprecision machining	2	50 nm	Large	High
	PMFE^［77］	Ultraprecision machining	2	600 nm	Large	High

Table 6. Measuring method of distance in feature point matching
View table
Table 6. Measuring method of distance in feature point matching
Method Expression
Euclidean distance $d (F, G) = \sqrt[]{\sum {(f_{i} - g_{i})}^{2}} (i = 1,2, \dots, n)$
Cosine similarity $c o s (F, G) = \frac{\sum (f_{i} \times g_{i})}{\sqrt[]{\sum a_{i}^{2}} \times \sqrt[]{\sum b_{i}^{2}}} (i = 1,2, \dots, n)$
Hamming distance $h (F, G) = \sum b_{i} (F) \otimes b_{i} (G) (i = 1,2, \dots, n)$

Tools

Get Citation

Copy Citation Text

Chenyang Zhao, Jie Xiang, Kai Bian, Zijian Zhu, Qinghong Wan. Review of Optical Microvision-Based Precision Positioning Measurement (Invited)[J]. Laser & Optoelectronics Progress, 2024, 61(2): 0211021

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites