Multi-frame sub-pixel processing algorithm based on uncontrolled micro-scanning

For the uncontrolled micro-scanning where the displacement of image sequence is random, a sub-pixel processing algorithm based on boundary recursive and subsequent error optimization is proposed in this paper. Neighboring pixels are used to approximate the boundary pixels of the original image in the algorithm. Then the approximate error is optimized by the gray statistical principle. At last a sub-pixel image with high resolution is reconstructed. The results show that our algorithm is better than the general over-sample reconstruction algorithms and robust in different displacement situations.     

Sub-pixel processing algorithm of the reducing boundary recursive error in staring FPA micro-scanning imaging

<正>By analyzing the error distribution rule of the boundary recursive reconstruction algorithm in controlled micro-scanning,a sub-pixel image processing algorithm is proposed to reduce the error.The gray statistical principle is used in the algorithm to optimize the error and acquire the sub-pixel image that approximates the original image.The simulation result shows that the effect of this algorithm is better than the oversample and simple boundary recursive algorithm(BRA),and it results in a good effect both in those of visible light and infrared imaging systems.Therefore,the application of this algorithm will enhance the performance of optoelectronic imaging systems.     

Sub-pixel processing for super-resolution scanning imaging system with fiber bundle coupling

A multilayer fiber bundle is used to couple the image in a remote sensing imaging system.The object image passes through all layers of the fiber bundle in micro-scanning mode.The malposition of adjacent layers arranged in a hexagonal pattern is at sub-pixel scale.Therefore,sub-pixel processing can be applied to improve the spatial resolution.The images coupled by the adjacent layer fibers are separated,and subsequently,the intermediate image is obtained by histogram matching based on one of the separated image called base image.Finally,the intermediate and base images are processed in the frequency domain.The malposition of the adjacent layer fiber is converted to the phase difference in Fourier transform.Considering the limited sensitivity of the experimental instruments and human sight,the image is set as a band-limited signal and the interpolation function of image fusion is found.The results indicate that a super-resolution image with ultra-high spatial resolution is obtained.     

Non-uniformity correction algorithm for IRFPA based on motion controllable micro-scanning and perimeter diaphragm strips

Non-uniformity correction is the key issue for the image quality improvement of infrared focal panel array (IRFPA) imaging. A non-uniformity correction (NUC) algorithm for IRFPA based on motion controllable micro-scanning platform and perimeter diaphragm strips is presented. We initially execute one-point calibration to the perimeter detectors, then based on controllable motion of adjacent frames, a special algebraic algorithm is proposed to transport the calibration of the perimeter detectors to those interior un-corrected ones. In this way, the bias parameter of the whole field of view (FOV) is calculated. The algorithm can be easily combined with sub-pixel imaging, thereby improving the quality of thermal imaging system (image spatial resolution and uniformity). All calculations are algebraic, with a low computation load. The algorithm can realize adaptive one point calibration without covering the central FOV rapidly. Experiments on simulated infrared data demonstrate that this algorithm requires only dozens of frames to obtain high quality corrections.     

基于Zernike正交矩的亚像素图像线宽测量算法

针对图像内的细窄线宽,提出了一种基于Zernike正交矩的亚像素图像线宽检测算法。该算法具有明确的几何模型,通过计算图像的2阶和4阶Zernike正交矩,推导出了亚像素线宽表达式。根据数字图像的离散性,给出了计算正交矩所需的模板系数,并分析了由离散性造成的原理误差。将所提出的亚像素线宽检测技术应用于安瓿内异物粒径的标定实验,结果表明:该算法可有效地测量弱小目标在图像中的亚像素线宽值,从而得到了异物粒径大小与亚像素线宽之间的标定曲线。     

Improving motion estimation by accounting for local image distortion

Cardiac elastography is a useful diagnostic technique for detection of heart function abnormalities, based on analysis of echocardiograms. The analysis of the regional heart motion allows assessing the extent of myocardial ischemia and infarction. In this paper, a new two-stage algorithm for cardiac motion estimation is proposed, where the data is taken from a sequence of 2D echocardiograms. The method combines the advantages of block-matching and optical flow techniques. The first stage employs a standard block-matching algorithm (sum of absolute differences) to provide a displacement estimate with accuracy of up to one pixel. At the second stage, this estimate is corrected by estimating the parameters of a local image transform within a test window. The parameters of the image transform are estimated in the least-square sense. In order to account for typical heart motions, like contraction/expansion, translation and rotation, a local affine model is assumed within the test window. The accuracy of the new algorithm is evaluated using a sequence of 500 grayscale B-mode images, which are generated as distorted, but known copies of an original ROI, taken from a real echocardiogram. The accuracy of the motion estimation is expressed in terms of errors: maximum absolute error, root-mean-square error, average error and standard deviation. The errors of the proposed algorithm are compared with these of the known block-matching technique with cross-correlation and interpolation in the sub-pixel space. Statistical analysis of the errors shows that the proposed algorithm provides more accurate estimates of the heart motion than the cross-correlation technique with interpolation in the sub-pixel space.     

Improving motion estimation by accounting for local image distortion

Cardiac elastography is a useful diagnostic technique for detection of heart function abnormalities, based on analysis of echocardiograms. The analysis of the regional heart motion allows assessing the extent of myocardial ischemia and infarction. In this paper, a new two-stage algorithm for cardiac motion estimation is proposed, where the data is taken from a sequence of 2D echocardiograms. The method combines the advantages of block-matching and optical flow techniques. The first stage employs a standard block-matching algorithm (sum of absolute differences) to provide a displacement estimate with accuracy of up to one pixel. At the second stage, this estimate is corrected by estimating the parameters of a local image transform within a test window. The parameters of the image transform are estimated in the least-square sense. In order to account for typical heart motions, like contraction/expansion, translation and rotation, a local affine model is assumed within the test window. The accuracy of the new algorithm is evaluated using a sequence of 500 grayscale B-mode images, which are generated as distorted, but known copies of an original ROI, taken from a real echocardiogram. The accuracy of the motion estimation is expressed in terms of errors: maximum absolute error, root-mean-square error, average error and standard deviation. The errors of the proposed algorithm are compared with these of the known block-matching technique with cross-correlation and interpolation in the sub-pixel space. Statistical analysis of the errors shows that the proposed algorithm provides more accurate estimates of the heart motion than the cross-correlation technique with interpolation in the sub-pixel space.     

基于相位相关的亚像素配准技术及其在电子稳像中的应用

介绍一种基于相位相关的图像配准方法,并将该方法应用于序列图像的运动检测,通过运动补偿,实现了序列图像的稳定输出。采用相位相关配准方法对序列图像进行亚像素精度配准,运用基于最小二乘的曲面拟合法,估计当前图像相对参考图像的亚像素级的平移量。在补偿图像运动时,为保证亚像素级的补偿精度,采用平滑算法,避免在图像补偿时出现马赛克现象。最后,对样本图像进行了亚像素级位移配准和图像补偿对比实验。实验结果表明,提出的方法可以检测到0．01pixel的运动量,最大配准误差为0．008pixel;采用亚像素级运动量补偿图像,最大误差〈0．5。     

Simple and fast rail wear measurement method based on structured light

In this paper, a fast and accurate rail wear measurement method based on simple equipments is presented. The inner rail profile is measured by a line structured light vision sensor. Using the centers of the big and small circle from the rail waist profile as control points, the measured rail profile is registered to the reference profile. The rail wear, including the vertical and horizontal rail wear, is computed by comparing the registered measured profile with the reference profile. The method has three key contributions: (1) the rail waist light stripe center points in the images are located fast and accurately by first tracking the region containing the rail waist light stripe using the Kalman filter and then computing the sub-pixel precision image coordinates by Hessian matrix at pixels. (2) The rail waist profile is segmented automatically into arcs of big and small circles by thresholding the normal angle curve of the measured rail waist profile. The centers of the two circles are used as control points for registering the measured rail profile to the reference profile. (3) The fast location of rail wear points in the images is realized by projecting the rail wear constraint points to the image, which simplifies the problem of computing rail wear from 2d image processing to 1d searching along the line segment connecting two rail wear constraint points. Experiments show that the proposed method can achieve 500 fps measurement frequency. At a train speed of 350 km/h, the interval between two consecutive measurements is about 190 mm. The system is tested on a real running train, and the measurement results are compared with those rail wear measured manually by special gage. The RMS errors of vertical and horizontal rail wears are 0.34 and 0.30 mm, respectively.     

Real-time blind deconvolution of retinal images in adaptive optics scanning laser ophthalmoscopy

With the use of adaptive optics (AO), the ocular aberrations can be compensated to get high-resolution image of living human retina. However, the wavefront correction is not perfect due to the wavefront measure error and hardware restrictions. Thus, it is necessary to use a deconvolution algorithm to recover the retinal images. In this paper, a blind deconvolution technique called Incremental Wiener filter is used to restore the adaptive optics confocal scanning laser ophthalmoscope (AOSLO) images. The point-spread function (PSF) measured by wavefront sensor is only used as an initial value of our algorithm. We also realize the Incremental Wiener filter on graphics processing unit (GPU) in real-time. When the image size is 512 × 480 pixels, six iterations of our algorithm only spend about 10 ms. Retinal blood vessels as well as cells in retinal images are restored by our algorithm, and the PSFs are also revised. Retinal images with and without adaptive optics are both restored. The results show that Incremental Wiener filter reduces the noises and improve the image quality.     

Simultaneous perimeter measurement for multiple planar objects

A novel scheme to measure the true boundary perimeters of multiple planar objects is proposed. This scheme is suitable for the simultaneous and accurate measurement of multiple objects’ boundary perimeters with different shape. A geodesic active contour is applied and this scheme consists of three steps. First, the active contour converges to the object's contour edge from an initial position via the iteration of the level set functional . Second, for the convergent active contour, the sub-pixel edge detection and the centroid measurement are applied to obtain every object's edge and centroid coordinate simultaneously. Finally, a centroid-self-calibration technology is applied, which measures the pixel's size equivalence with the computed centroid and a dual-frequency laser to measure every object's true boundary perimeter. Experiments indicate that this scheme's measurement repetition error decreases to ±0.4% and that its single measurement error decreases to ±0.2%. Compared to conventional perimeter measurement schemes, this scheme is highly efficient (i.e., simultaneous measurement of multiple objects’ boundary perimeters) and highly accurate.     

Digital image correlation for large deformation applied in Ti alloy compression and tension test

In this paper, digital speckle correlation is used in the measurement of Ti alloy compression and tension test. The key technologies applied in the measurement are discussed in detail, including camera calibration with telephoto lens and digital image correlation in large deformation. Single camera self-calibration algorithm based on photogrammetry is proposed. In the algorithm, the interior parameters of camera are estimated without calibrated object, using the bundle adjustment technique, so the 3-D coordinates of calibration target points are not needed in advance to get a reliable camera calibration result. An updating reference image scheme could be employed to deal with large deformation situation. A large deformation measurement scheme, updating reference image scheme, is proposed in this paper. The un-deformed image is used as reference in correlation at first. Only for extremely large deformation field, in which iteration of correlation is not convergent, the reference image is updated to the image of previous deformed stage. Using this method, not only extremely large deformation can be measured successfully but also the accumulated error could be controlled. The 75 mm lens is calibrated in the measurement and compared the result with extensometer and un-calibrated image. Experimental results show that up to 150% tensile deformation and 50% compression deformation can be measured successfully.     

Estimation of speed of sound in dual-layered media using medical ultrasound image deconvolution

The speed of sound in soft tissues is usually assumed to be 1540 m/s in medical pulse-echo ultrasound imaging systems. When the true speed is different, the mismatch can lead to distortions in the acquired images, and so reduce their clinical value. Previously we reported a new method of sound-speed estimation in the context of image deconvolution. Unlike most other sound-speed estimation methods, this enables the use of unmodified ultrasound machines and a normal scanning pattern. Our approach was validated for largely homogeneous media with single sound speeds. In this article, we demonstrate that sound speeds of dual-layered media can also be estimated through image deconvolution. An ultrasound simulator has been developed for layered media assuming that, for moderate speed differences, the reflection at the interface may be neglected. We have applied our dual-layer algorithm to simulations and in vitro phantoms. The speed of the top layer is estimated by our aforesaid method for homogeneous media. Then, when the layer boundary position is known, a series of deconvolutions are carried out with dual-layered point-spread functions having different lower-layer speeds. The best restoration is selected using a correlation metric. The error level (e.g., a mean error of −9 m/s with a standard deviation of 16 m/s) for in vitro phantoms is found to be not as good as that of our single-speed algorithm, but is comparable to other local speed estimation methods where the data acquisition may not be as simple as in our proposed method.     

A novel algorithm in buildings/shadow detection based on Harris detector

A novel method based on corner detectors is proposed in detecting shadow and buildings in this paper. Its most outstanding point is employing Harris corner detector in region-based detection, despite that Harris detector traditionally used to select pixels as final results. Different densities of buildings are generally influenced by different features for recognition. First time, images are self-grouped into two groups according to the distribution of buildings, and two specifical algorithms are ready for detection specifically. A region-based method is used in comparison with our algorithm, and the results indicate that the new idea works not only more robustly, but also more effectively. It is a fast and simple method, which needs average 3.28 × 10⁻⁵ s to run per square image.     

Shooting algorithm and particle swarm optimization based Raman fiber amplifiers gain spectra design

An initial value determination method with a contraction factor for the counter-pumped Raman coupled equations is proposed. This method is used in conjunction with initial guess correction mechanism of Newton's method to construct a new efficient shooting algorithm for the solution of counter-pumped Raman coupled equations. The particle'swarm optimization is used to find the optimal wavelengths and powers for the pumps. By combining the new shooting algorithm and particle swarm optimization a powerful approach to the design of gain spectra for Raman fiber amplifiers is developed. Using this approach a counter-pumped broadband Raman fiber amplifier in C + L-band is designed and optimized. An average on-off gain of 9.3 dB for a bandwidth of 95 nm is obtained using only 4 pumps, with an in-band ripple level of ± 0.7 dB.     

基于Powell算法的亚像元配准方法

基于Powell优化算法和图像的金字塔分解,提出了一个新的亚像元运动参数估计方法.用参考图像和待配准图像构造的非线性最小二乘公式作为Powell算法的目标函数,并且以运动参数作为变量.由细到粗的多分辨率分解为准确的亚像元参数估计提供了基础.数值实验结果表明,该算法可以用于准确的稳健的亚像元参数估计,其不足之处在于运行时间比较长.     

一种基于相位相关的亚像素红外图像配准算法

图像匹配技术广泛应用于各种图像处理任务,如图像拼接、机器视觉等。通常匹配算法的精确度只能达到像素级别,但在很多图像处理任务如超分辨率重建中需要亚像素精度的图像配准。提出了一种基于相位相关的亚像素图像配准算法。根据两幅离散数字图像的相位相关矩阵中的最大值以及其附近若干点可以拟合估计出实际的峰值位置,进而实现两幅图像的亚像素运动估计。提出的算法针对热像仪采集的红外图像进行匹配实验,实验结果表明该算法精度相比通常的亚像素匹配算法较高,且具有更好的实用性。     

The high precision positioning algorithm of circular landmark center in visual measurement

For vision measurement in the center circle landmark location problem, an improved center of the ellipse based on Zernike moments sub-pixel positioning algorithm was proposed. First, using Sobel operator edge of the image pixel level positioning and then use the constructed Zernike moments to solve the model, combined with Zernike orthogonal polynomials and completeness and plural moment magnitude rotational invariance calculated edge sub-pixel position; followed by analysis of the principle deviation generated by the moments template and ideals model, a correction formula was proposed to compensate and improved edge criterion used to image sub-pixel edge positioning; finally, using the least-squares ellipse fitting algorithm fitting circle center, reverse edge point, filtered residuals larger point, and then precise positioning of the ellipse center. The experimental results show that this method has good stability and high positioning accuracy can be efficiently used in many applications.     

一种基于边缘特征的亚像素投影配准算法

为了提高红外图像配准准确度,充分利用图像的边缘信息,结合线性插值,对原有的投影算法引入新的图像相似性评价函数——均方误差函数(Mean Sequare Erron,MSE),提出了基于边缘特征的亚像素投影配准算法.提取图像边缘特征明显的部分做配准能提高匹配概率,引入积分投影能有效降低算法的计算复杂度.算法中还运用了线性插值,使配准准确度达到亚像素级水平.     

Two-dimensional Otsu's thresholding segmentation method based on grid box filter

Otsu algorithm, an automatic thresholding method, is widely used in classic image segmentation applications. In this paper, a novel two-dimensional (2D) Otsu thresholding algorithm based on local grid box filter is proposed. In our method, firstly by utilizing the coarse-to-fine idea, the 2D histogram is divided into regions by grid technique, and each region is used as a point to form a new 2D histogram, to which 2D Otsu thresholding algorithm and an improved particle swarm optimization (PSO) algorithm are applied to get the region number of the new 2D histogram threshold. Then on the result region, the mean of the 2D histogram is computed base on box filter, and the two algorithms are applied again to obtain the final threshold for the original image. Experimental results on real data show that the proposed algorithm gets better segmentation results than the traditional recursion Otsu algorithm. It significantly reduces the time of segmentation process and simultaneously has the higher segmentation accuracy.