With the development of high-generation production lines for Liquid Crystal Display (LCD) panel, online real-time detection of large-scale, high-resolution LCD surface defect imposes stringent requirements on image algorithms and computing platforms. This paper aims at the edge effect in the traditional one-dimensional Fourier reconstruction LCD detection algorithm. Based on the potential periodicity of the Fourier transform and the approximate periodic arrangement of the element array in LCD panel images, a complete period truncation image preprocessing strategy is proposed. The processing strategy not only eliminates the ring effect at the connection joint caused by the original algorithm's extending period at both ends of the image, but also reduces the length of the Fourier operation, which is even more noticeable when detecting large-size panels. In addition, in order to adapt to the fast Fourier transform input interface, the resampling method is applied to the complete period truncated image whose length is not an integer power of 2, and an FPGA-based resampling acceleration structure is designed. The experimental results show that compared with the traditional one-dimensional Fourier reconstruction algorithm, our proposed method not only performs better in eliminating LCD global periodic texture, but also reduces the amount of data to be processed. The FPGA acceleration scheme even reduces the scanning detection time of a 8.5-generation LCD panel to 12.5s
The surface laser speckle image is obtained by the reflected and scattered light beams from a rough surface illuminated by laser. Based on the fractal theory, Double Blanket Model (DBM) is proposed to analyze laser speckle images. The dimension of the space surface is regarded as the characteristic parameter in DBM method. Laser speckle images are preprocessed to remove interference and noise from the environment at first. The size and direction of optimum window are researched. The DBM characteristic parameter is calculated under the optimum window. The relationships are researched between DBM characteristic parameter and surface roughness R<sub>a</sub>. The results show that the surface roughness contained in the surface speckle images has a good monotonic relationship with DBM characteristic parameter. To obtain roughness value through a laser speckle image, the fitting function relationship between R<sub>a</sub> and DBM characteristic parameter is established, and the fitting function stability is analyzed by experiments. The experiment results show that surface roughness measurement based on DBM method of laser speckle is feasible and applicable to on-line high-precision roughness detection, which has some advantages such as non-contact, high accuracy, fast, remote measurement and simple equipment.
We design a directional phase-shift circular arrays targets to calibrate camera system under the assumption that targets are in a defocus condition. In contrast to conventional camera calibration method, we can achieve accurate camera calibration with severely defocused images by using the proposed directional phase-shift circular arrays targets. In this paper, three directional phase-shift circular arrays targets are used to calibrate the measurement system. We set up a mathematical model to detect the feature points accurately even though the targets images are seriously blurred. In the experiments, the targets are displayed in a LCD monitor, camera system can capture the three targets images at the same position to extract the feature points. Gaussian filters are utilized to blur the targets images to simulate the fuzzy images. And we obtain targets images with different defocus degree by adjusting the depth range of the actual camera system. We have done a series comparative tests with checkerboard. We evaluate the performance of the proposed method on simulated and real data. Finally, the measurements of displacement and morphology are implemented. Experimental results indicate that the directional phase-shift circular arrays targets can be used to achieve accurate camera calibration parameters.
This paper describes a method of non-contact distance measurement in high temperature environments by video extensometer. In order to detect the distance of a pair of marked lines on the surface of the specimen, dual fitting algorithms are proposed. The first fitting method which is least squares curve fitting utilizes hyperbolic tangent function to fit a serial of discrete gray level points to a curve. So a set of edge points on the marked line was located. The second is to use orthogonal linear least squares method to fit two sets of edge points to two lines respectively. After the two lines are located, the distance of two lines can be computed. A video extensometer system which enables a high temperature distance measurement up to 1200°C is presented and demonstrated by experimental measurements.
Due to high-precision, self-cleaning, anti-interference, non-contact and the other advantages, pneumatic measuring has been widely used in the high-precision, non-contact measurement. Traditional back-pressure measurement had been used in ever cases, but with small measuring range (200μm), its application is limited. While reflection-type pneumatic sensor can be used for online, non-contact measurement, and with large measuring rang (4~5mm) and high-precision (5~10μm), it may also be applied to the sensor calibration. But structural design of its measuring head has a huge impact on the range and accuracy of the sensor. In this paper, the operation principle of a new large range non-contact dimension measuring device is introduce, based on hydrodynamics, a finite element model of the measuring head is systematically constructed, by using finite element simulation software, the influence of sensitivity and linear rang from multiple structural parameters and flow parameters would be analyzed. Through simulation results under different parameters, we obtained optimal geometrical parameters of reflection-type pneumatic head, so as to achieve the purpose of optimizing the design of reflection-type pneumatic head. From the simulation results, the characteristic curve after optimization has good linearity and large measurement range.
With respect to orifice throttling or compensating, capillary throttling has following advantages: smaller mass flow rate and stronger anti-interference ability. This paper firstly gives the required average pressure of air-film when shipping a piece of LCD glass. Then, dimensional flow model of the capillary throttling of air-flotation unit is established. Based on the model, we firstly analyze the flowing process of the lubricated air through the capillary. Secondly, the pressure distribution equation of air-film is derived from the Navier−Stokes Equation. Furthermore, the approximate functional relations between model parameters and static characteristics of the air-film, such as mass flow rate, static bearing capacity, are obtained and then influence of the former on the latter is analyzed . Finally, according to the continuity of air flow, the function relation between model parameters and pressure of core nodes in the air-film is also derived. On foundation of theoretical analysis, the impacts of each model parameter on static characteristics of the air-film flow field, are respectively simulated and analyzed by CFD software Fluent. Based on these simulations and analysis, radius and length of the capillary, density of the gas supply orifices and other model parameters are optimized. Finally, the best unit model is acquired, which greatly improves the static working performance of air-film in air-flotation unit. Research results of this paper can provide guidance and basis for the design and optimization of air-flotation transporting system.
Flat panel displays have been used in a wide range of electronic devices. The defects on their surfaces are an important factor affecting the product quality. Automated optical inspection (AOI) method is an important and effective means to perform the surface defection inspection. In this paper, a kind of defect extraction algorithm based on one dimensional (1D) Fourier theory for the surface defect extraction with periodic texture background is introduced. In the algorithm, the scanned surface images are firstly transformed from time domain to frequency domain by 1D Fourier transform. The periodic texture background on the surface is then removed by using filtering methods in the frequency domain. Then, a dual-threshold statistical control method is applied to separate the defects from the surface background. Traditional 1D Fourier transform scheme for detecting ordinary defects is very effective; however, the method is not where the defect direction is close to horizontal in periodic texture background. In order to tackle the problem, a mean threshold method based on faultless image is put forward. It firstly calculates the upper and lower control limits of the every reconstructed line scanned image with faultless and then computes the averages of the upper and lower limits. The averages then act as the constant double thresholds to extract the defects. The experimental results of different defects show that the method developed in the paper is very effective for TFT-LCD panel surface defect inspection even in the circumstance that the defect directions are close to horizontal.
A vision-based on-machine measurement system (OMM) was developed to improve manufacturing effectiveness. It was based on a visual probe to enable the CNC machine tool itself to act as a coordinate measuring machine (CMM) to inspect a workpiece. The proposed OMM system was composed of a visual probe and two software modules: computer-aided inspection planning (CAIP) module and measurement data processing (MDP) module. The auto-focus function of the visual probe was realized by using astigmatic method. The CAIP module was developed based on a CAD development platform with Open CASCADE as its kernel. The MDP module includes some algorithms for determination of inspection parameters, for example, the chamfered hole was measured through focus variation. The entire system was consequently verified on a CNC milling machine.
A non-contact surface roughness measurement technique is put forward based on texture analysis of the digital laser
speckle pattern coming from a measured grinding surface. The speckle pattern was captured by a digital camera while
the grinding metal surface was illuminated by a laser. Then the surface roughness information immerged in the high
frequency sub-band of the speckle pattern was extracted with the texture analysis of wavelet transform and Markov
Random Fields model. Our research has illustrated that surface roughness Ra has a good monotonic relationship with the
texture features of the speckle pattern. If this measuring system is calibrated with the surface standard samples roughness
beforehand, the surface roughness actual value Ra may be obtained in the case of the same material surfaces ground at
the same manufacture conditions.
A polarization-sensitive white light interferometer (PSWLI), which is a promising technique that can be used to measure the stress induced birefringence, is developed. The use of wide-spectrum light source brings to PSWLI a resolution in the micron range, but the difficulty arises when searching for the best fringe contrast within the extremely short coherence length, especially for the Linnik interference configuration. To tackle this problem, an autofocus device based on the improved astigmatic method is embedded in the PSWLI system to firstly automatically determine the best foci of the reference mirror and the test sample. Then the minimization of optical path difference (OPD) of two interference arms are implemented by the root mean square fringe contrast function combined with a 4×4 pixel binning of the CCD camera. The autofocus time is no more than 0.3 second and the minimization of OPD has a speed of 2.2 min/mm. Finally, the developed PSWLI system is calibrated by the Berek compensator and the birefringence measurement result is obtained by simultaneously calculating the phase retardation, the optical axis orientation, the reflectance and the stress map of a stress induced birefringence sample.
Computer-Aided Inspection Planning (CAIP) is an important module of modern dimensional measuring instruments,
utilizing the CAIP for machined parts inspection is an important indication of the level of automation and intelligence.
Aiming at the characteristic of visual inspection, it develops a CAIP system for vision-based On-Machine Measurement
(OMM) based on a CAD development platform whose kernel is Open CASCADE. The working principle of
vision-based OMM system is introduced, and the key technologies of CAIP include inspection information extraction,
sampling strategy, inspection path planning, inspection codes generation, inspection procedure verification, data
post-processor, comparison, and so on. The entire system was verified on a CNC milling machine, and relevant examples
show that the system can accomplish automatic inspection planning task for common parts efficiently.
Uniform irradiance and color adjustability are the key features in the design of lighting for machine vision inspection systems. A simple and practical design tool of angled light-emitting-diode (LED) ring arrays for uniform near-field irradiance has been developed by introducing a simple model to simplify the complexity of nonrotational symmetric irradiance distribution of angled LEDs. The color distribution and color uniformity of the ring array assembled with RGB LEDs are analyzed based on the analytical model of color mixing. According to the theoretical analysis, the simulated results, and the design exemplifications, the practical design tool offers an easy way to estimate the performance of an RGB LED ring array and can be considered as a starting point to reduce the computation time for exact designs that must use a realistic LED model.
A tracking measurement method is introduced using pan-tilt-zoom cameras based on machine vision theory. The
modeling of a pan-tilt-zoom camera system is described. The model formula given represents the relationship between
the coordinates of a spatial point and its corresponding image coordinates when the system focus, zoom, aperture and
illumination are adjusted to track the measured object. The modeling technique has excellent applicability and facilitates
the assembly and alignment of the camera. The only requirement for alignment is to align the image sensor plane parallel
to the tilt rotation axis.
In order to improve the measurement speed of a large scale multi-sensor visual measurement system, in this paper a novel high speed measurement system based on 1394 bus is introduced. The implementation of 1394 bus in the system make both the image data from the visual sensors and the controlling signals can share the same high speed 1394 bus. This can overcome some drawbacks, such that the data and controlling signals transmit in data bus and controlling bus separately, which are often used in traditional large scale multi-sensor visual measurement systems. Another merit of using 1394 bus in the system is to make the system more compact, reliable and maintainable. In the system, the parallel image grabbing and processing units are also adopted. Consequently, the measurement time of the system can be reduced dramatically for real-time measurement situations. It is also worth noting that a flash programmable memory is used in each image processing unit for facilitating the measurement system to measure different objects.
Considering the potential clinical importance, the surface tension of ocular cornea under the action of normal physiological intraocular pressure is estimated, and a novel technique and a simple mechanical model for determining the tension are also presented in this paper. An instrument embodying mainly a CCD camera, an optical staff gauge and a manometer was developed primarily to measure both the surface point displacement and intraocular pressure of the cornea. A simple theoretical model was used to characterize the tensions of the ocular corneas under the action of the intraocular pressure. Due to the difficulty in obtaining the human cornea, laboratory experiments were carried out on porcine cornea specimens. The thickness of the specimens was accurately measured by optical coherence tomography. The matrix and collagen properties within the corneal tissue were manifested in the experiment. Experimental results on porcine corneas showed that the present technique is applicable to estimate the surface tension. In the normal physiological intraocular pressure range, both meridian and circumference tensions of the porcine corneas along the radial coordinate distribute are not uniform.
A new self-measurement tonometer operated easily by the patient at home is presented in this paper. Based on the total
reflection principle, a cone prism was designed as a measuring probe and associated a photoelectric transducer acting as
applanation area detector converted the diminished quantity of light returned from applanation surface into one
electronic signal, and a dial acting as indicator converted the applanation force of the eye into intraocular pressure. A
central processing unit with the read-only memory stored the program instructions and the interrelated data. Calibration
experiments were carried out on a stimulated cornea clamped in a Perspex chamber connected to a hydraulic manometer
to obtain intraocular pressure at different levels. Preliminary clinical testing was carried out comparing the values
obtained with those of the Goldmann tonometer. Experimental results demonstrate that the present tonometer reading has
good agreement with that of the Goldmann tonometer, which can meet patient's needs.
In this paper, the statistical properties of speckle pattern are investigated from the point of view of computer texture analysis, and a measurement method of an object surface roughness and the object surface displacement is subsequently put forward. In the geometrical information extraction technique, the speckle pattern images are taken by a very simple configuration of setup consisting of a laser and a CCD camera. Surface roughness and displacement information is extracted based on the energy feature characterization of the grey level co-occurrence matrices of the speckle pattern images. Our experimental results show that surface roughness and displacement contained in surface speckle pattern images have a good relationship with their energy feature. By means of a set of standard surface specimens with different roughness, the relation curve is readily calibrated. Then the surface roughness of an object surface made by same material and manufactured by the same way as the standard specimens can be evaluated from a single speckle pattern image taken from the surface. Similarly, the surface displacement can also be obtained from a single speckle pattern image. The easily implemented method can be used in-process surface roughness and displacement measurement.
A surface scanner has been developed which records the light from an approximately flat surface at up to sixteen different angles. The scanner simultaneously records images of the surface as an illuminated spot is traversed across the surface. The optical design of the scanner allows the use of just one parabolic mirror slightly larger than the scan length required, and small detectors for the reflected/scattered light at each angle. Design principles for the scanner are developed and some resulting images on surfaces presented. The data from the above scanner is used to compute the surface topology of a surface by integrating the surface slope measurements obtained from the multi-angle data. This is accomplished by obtaining an estimation of surface slope at each pixel position. Constraints are then imposed on the data to obtain a continuous three-dimensional surface. The surface obtained is compared with data obtained from the surface by measuring with a "Talysurf" stylus measurement machine. To do this a common coordinate frame is needed requiring a rotation and scaling in two directions of one of the data sets to obtain maximum cross-correlation between the three dimensional profiles.
A new finite element procedure for the solution of the weight function, which represents the degree of the contribution of
the fluid velocity to the signal in the cross section of a pipe of electromagnetic flow meter, at every point of the cross
section of flow pipe of multiphase flow was presented. The solution of the weight function is important for the velocity
profile inversion process. First, a numerical simulation model was built up with Femlab. Then a comparison study was
investigated using the numerical simulation and Shercliff's weight function. It proved that the finite element
methodology is correct for solving the weight function. Based on this result, a new kind of weight quantity solution
scheme was carried out when the electrodes are located on the internal circumference of flow pipe and the connecting
line of electrodes is parallel to the pipe diameter that is superposition with the coordinate axis. This methodology is
effective to obtain the weight function value of electromagnetic flow meter and will be used in the invert reconstruction
of the velocity profile imaging, which is significantly important for volumetric flow rate measurement.