An image pixels based fast matching algorithm for Die Bonder targeting

Prior research studies have shown that the peak signal to noise ratio (PSNR) is the most frequent watermarked image quality metric that is used for determining the levels of strength and weakness of watermarking algorithms. Conversely, normalised cross correlation (NCC) is the most common metric used after attacks were applied to a watermarked image to verify the strength of the algorithm used. Many researchers have used these approaches to evaluate their algorithms. These strategies have been used for a long time, however, which unfortunately limits the value of PSNR and NCC in reflecting the strength and weakness of the watermarking algorithms. This paper considers this issue to determine the threshold values of these two parameters in reflecting the amount of strength and weakness of the watermarking algorithms. We used our novel watermarking technique for embedding four watermarks in intermediate significant bits (ISB) of six image files one-by-one through replacing the image pixels with new pixels and, at the same time, keeping the new pixels very close to the original pixels. This approach gains an improved robustness based on the PSNR and NCC values that were gathered. A neural network model was built that uses the image quality metrics (PSNR and NCC) values obtained from the watermarking of six grey-scale images that use ISB as the desired output and that are trained for each watermarked image’s PSNR and NCC. The neural network predicts the watermarked image’s PSNR together with NCC after the attacks when a portion of the output of the same or different types of image quality metrics (PSNR and NCC) are obtained. The results indicate that the NCC metric fluctuates before the PSNR values deteriorate.

In this paper we propose a fast normalized cross correlation (NCC) algorithm for pattern matching based on combining adaptive multilevel partition with the winner update scheme. This winner update scheme is applied in conjunction with an upper bound for the cross correlation derived from Cauchy-Schwarz inequality. To apply the winner update scheme, we partition the summation of cross correlation into different levels with the partition order determined by the gradient energies of the partitioned regions in the template. Thus, this winner update scheme can be employed to skip the unnecessary calculation. Experimental results show the proposed algorithm is very efficient for image matching under different lighting conditions.

A stereovision algorithm is proposed for visual odometry to estimate motion of mobile robot by providing feature pair sequence. It is composed of feature extracting, matching and tracking. Firstly, corners are extracted as features by Harris operator and grid-based optimizing. In feature matching and tracking, serious problems are caused by variable illumination between stereo images. An improved Moravec's Normalized Cross Correlation (MNCC) algorithm is presented to reduce illumination affect in computing correspondence of corners. On current stereo image pair, extracted corners are matched by correlation-based bidirectional algorithm and outliers are rejected by epipolar constraint. Matched corners are tracked in pre-estimated search windows. The computational cost is greatly reduced by limiting number of corners, pre-estimating search window and feature local-updating. Simulation results validate that our algorithm is efficient and reliable.

Extraction of micro-vibration information from a single-shot speckle image is realized base on the row-by-row exposure mechanism of a rolling shutter camera and the high intensity of laser speckle. Normalized cross correlation and cubic spline interpolation algorithm is firstly used between nonoverlapping gray scales of adjacent rows to infer micro-displacement information. With number of rows set as 342, wide frequency range from 300 Hz to 20 kHz sinusoid signals are recovered with relative errors smaller than 2%. For low frequency signals from 40 Hz to 300 Hz, the relative errors of the extracted frequency are smaller than 5%. The effects of number of sampling cycles (NSC) and sampling points in unit period on the system performance are investigated and high accuracy is obtained with NSC of 4 and sampling points of 3 per signal period. In a single shot image, spatial multi points vibration detection scheme is presented and experimentally realized. The representative signals of 5 kHz, 10 kHz and 20 kHz are successfully extracted in the multidimensional scenes of 1D, 2D and 3D respectively, and the relative errors are all within 0.5%. This scheme can be potentially used in air surveillance for wide range remote sensing of interested object or sounding.

Image matching algorithms based on image gray value are commonly used, which can achieve high matching accuracy. Among them, the Normalized Cross Correlation (NCC) method has high accuracy and strong adaptability, however it has the disadvantages of high computational complexity and slow calculation speed. By using dynamic programming strategy, a fast template matching algorithm based on NCC is proposed in this paper. At the same time, by calculating the cumulative sum of pixel values and the sum of squares of pixel values, and using the down sampling process, the calculation time and complexity are greatly reduced. Experimental results show that the proposed method has the same matching accuracy as the existing algorithms, and further improves the image matching efficiency.

In this paper, we introduce a time varying smoothing factor for the noise update in the optimal filtering based on QR decomposition (QRD). Conventionally, the noise statistics are updated during noise-only periods and kept fixed during speech present periods in QRD-based multichannel noise reduction [G. Rombouts and M. Moonen, Sept 2003]. Instead of using this conventional scheme that needs an explicit voice activity detector (VAD), we adopt a time varying smoothing factor which is parameterized from the normalized cross correlation (NCC). In a multi-microphone based system, the NCCs between two microphones can be used for finding the direction of signals [M. Omologo and P. Svaizer, 1994]. Provided that we know the direction of the speech signal and the time delays between microphones are compensated, the NCCs can be considered as a measure for the speech presence probability. The NCCs are estimated for a given time interval, and they are averaged over the microphones and smoothed in time. Then the averaged value is mapped to a smoothing factor through a nonlinear function. The simulation results show that the proposed algorithm yields better performance compared to the conventional QRD-based noise reduction.

Sub-pixel image registration on an embedded Nanosatellite Platform

We describe the automated application of an area based registration method to the surface inspection of steel industry products as a tool to solve an intermediate mosaicing problem. The main problem of area based methods is that there is high probability that the results of a matching process will be incorrect if a region of interest without any relevant detail is used. The selection of a region of interest with relevant content continues to be a problem nowadays. We propose a method to select a salient area when using a zero mean normalised cross correlation metric and a block as a region of interest. The selection of the size and the position of the block is focused on ensuring a smooth unimodal similarity surface around the maximum similitude point. Experiments show a correlation between the surface kurtosis of the block autocovariance and the same coefficient measured over the correlation surface around the maximum similitude point for the three different steel products analysed. We check that the maximum correlation value is reached abruptly, in a small range of pixels around the maximum similitude point, in correlation surfaces obtained from blocks containing non-relevant information. On the other hand, salient blocks usually lead to unimodal smooth similarity surfaces with small sensitivity to noise in contrast with the ones obtained from non-remarkable blocks. Also, the method proposed allows the application of fast search algorithms based on the unimodality of the correlation surface, obtaining high computational time reduction in comparison with full search strategies using fast normalised cross correlation algorithms.

Purpose: To develop and evaluate a template matching algorithm for direct detection of a lung tumor trajectory in the raw projections acquired from a cone beam CT. Method and Materials: A planning respiration‐correlated CT (pRCCT) was used to construct digitally‐reconstructed radiographic (DRR) templates for registration with raw projection data of a free‐breathing cone beam CT (FB‐CBCT). The pRCCT and FB‐CBCT were first registered to obtain the approximate mean 3D position of the tumor. This position was applied as a shift to the planning isocenter in the pRCCT. Next, 72 DRRs were generated at 5 degree angular increments around this shifted position, for each phase in the pRCCT. The appropriate phase DRR template for registration was selected using the measured phase of the FB‐CBCT projection. This phase was measured by tracking the diaphragm apex position in the projections. Two template matching algorithms were compared: normalized cross correlation and a block normalized cross correlation. A dynamic phantom was induced to move reproducibly 2cm during planning and 3cm during online CBCT imaging. The phantom was imaged at isocenter and with a 1cm translational shift from isocenter. Results: The tumor excursion measured with the normalized cross correlation method was 28.7 mm, and 28.5 mm with the block normalized cross correlation. The error in detecting the daily standard deviation of tumor position was 0.1mm for both methods. The mean absolute difference in detected tumor position was 1.07 mm, with a 95% confidence value of 2.43 mm. Block normalized cross correlation reduced the maximum error in tumor position detection slightly, in relation to normalized cross correlation. Conclusion: Direct detection of the tumor position from cone beam CT projections is feasible with template matching, provided that appropriate and accurate templates are used. Conflict of Interest: Supported in part by NIH R01 CA116249 and Elekta, Inc.

Featured with high anti-noise ability and precision, normalized cross-correlation (NCC) algorithm is broadly used in the feature point matching. However, it's not applied to the real-time processing because of the complexity of searching procedure and the great calculation quantity. To solve these problems, an improved NCC matching procedure is provided in this paper. First, the direction parameters of the feature points are abstracted with the histogram of gradient orientation (HOG), the pairs of the feature point in different gradient direction are filtered, and the matching matrix is used to get the right matching pair in order to avoid the repeated searching procedure. It can be included from the experiment result that the algorithm is not only characterized with good matching efficiency, but also can improve the calculation speed, which is good for the real-time processing and image matching with small amount of rotation.

The task of matching image patches is a fundamental problem in computer vision. When sufficiently textured patches are normalized up to similarity transformation, a simple Normalized Cross Correlation (NCC) of corresponding patches will give a high value. In practice, using it on patches per se may not perform well due to the noisy variations of pixel intensities. A more prudent approach will be to apply it to the abstract features extracted by a deep convolutional network. We study the applicability of an NCC based convolutional network for the task of Patch Matching. Further, there may be cases where the network may fail due to insufficient textures. In those cases, a simple pixel difference based method will be beneficial. To this end, we propose to improve the two basic architectures, Siamese networks and Central-Surround stream networks, using robust matching layers for learning the similarities of patches, assisted by a simple cross-entropy loss function. We empirically verify the performance of the proposed models on the challenging UBC Patches dataset and show that they are close to the state-of-the-art. Further, we evaluate their resilience to large illumination changes in two experimental scenarios: 1) by manually varying the patches of UBC Patches by an affine model 2) by using the publicly available Webcam dataset. We demonstrate that our models are indeed very resilient to illumination variations; they reduce the false positive rate to nearly 10%, and improve over the popular methods by nearly 5%. Further, we demonstrate the generalisability of the proposed NCC based matching layer by applying it to Face Recognition and show that it improves the performances of well known networks on a real-world, surveillance dataset.

As the technology for simultaneously recording from many brain locations becomes more available, more and more laboratories are measuring the cross-correlation between single-neuron spike trains, and between composite spike trains derived from several undiscriminated cells recorded on a single electrode (multiunit clusters). The relationship between single-unit correlations and multiunit cluster correlations has not yet been fully explored. We calculated the normalized cross-correlation (NCC) between single-unit spike trains and between small clusters of units recorded in the rat somatosensory cortex. The NCC between small clusters of units was larger than the NCC between single units. To understand this result, we investigated the scaling of the NCC with the number of units in a cluster. Multiunit cross-correlation can be a more sensitive detector of neuronal relationship than single-unit cross-correlation. However, changes in multiunit cross-correlation are difficult to interpret uniquely because they depend on the number of cells recorded on each electrode and because they can arise from changes in the correlation between cells recorded on a single electrode or from changes in the correlation between cells recorded on two electrodes.

Recently, Template matching approach has been widely used to locate faces with various pose, illumination and clutter background. Normalized Cross-correlation (NCC) is an effective and simple measurement method to compute the similarity matching between the stored faces templates and the rectangular blocks of the input image to locate the face position. However, localization error occurs very often due to some rectangular blocks which have more face than correct blocks because of the effect of matrices values of these blocks. In this paper we proposed a simple preprocessing method before the use of NCC. This is to reduce the effects of such problems by increasing the values of the input image pixels. The result showed a significant improvement in localization accuracy compared with the use of NCC alone which is only up to 11%. Yale University database was used to evaluate our proposed method.

Task automation is essential for efficient and consistent image segmentation in radiation oncology. We report on a deep learning architecture, comprising a U-Net and a variational autoencoder (VAE) for automatic contouring of the prostate gland incorporating interobserver variation for radiotherapy treatment planning. The U-Net/VAE generates an ensemble set of segmentations for each image CT slice. A novel outlier mitigation (OM) technique was implemented to enhance the model segmentation accuracy. The primary source dataset (source_prim) consisted of 19 200 CT slices (from 300 patient planning CT image datasets) with manually contoured prostate glands. A smaller secondary source dataset (source_sec) comprised 640 CT slices (from 10 patient CT datasets), where prostate glands were segmented by 5 independent physicians on each dataset to account for interobserver variability. Data augmentation via random rotation (<5 degrees), cropping, and horizontal flipping was applied to each dataset to increase sample size by a factor of 100. A probabilistic hierarchical U-Net with VAE was implemented and pretrained using the augmented source_prim dataset for 30 epochs. Model parameters of the U-Net/VAE were fine-tuned using the augmented source_sec dataset for 100 epochs. After the first round of training, outlier contours in the training dataset were automatically detected and replaced by the most accurate contours (based on Dice similarity coefficient, DSC) generated by the model. The U-Net/OM-VAE was retrained using the revised training dataset. Metrics for comparison included DSC, Hausdorff distance (HD, mm), normalized cross-correlation (NCC) coefficient, and center-of-mass (COM) distance (mm). Results for U-Net/OM-VAE with outliers replaced in the training dataset versus U-Net/VAE without OM were as follows: DSC=0.82±0.01 versus 0.80±0.02 (p=0.019), HD=9.18±1.22 versus 10.18±1.35mm (p=0.043), NCC=0.59±0.07 versus 0.62±0.06, and COM=3.36±0.81 versus 4.77±0.96mm over the average of 15 contours. For the average of 15 highest accuracy contours, values were as follows: DSC=0.90±0.02 versus 0.85±0.02, HD=5.47±0.02 versus 7.54±1.36mm, and COM=1.03±0.58 versus 1.46±0.68mm (p<0.03 for all metrics). Results for the U-Net/OM-VAE with outliers removed were as follows: DSC=0.78±0.01, HD=10.65±1.95mm, NCC=0.46±0.10, COM=4.17±0.79mm for the average of 15 contours, and DSC=0.88±0.02, HD=7.00±1.17mm, COM=1.58±0.63mm for the average of 15 highest accuracy contours. All metrics for U-Net/VAE trained on the source_prim and source_sec datasets via pretraining, followed by fine-tuning, show statistically significant improvement over that trained on the source_sec dataset only. Finally, all metrics for U-Net/VAE with or without OM showed statistically significant improvement over those for the standard U-Net. A VAE combined with a hierarchical U-Net and an OM strategy (U-Net/OM-VAE) demonstrates promise toward capturing interobserver variability and produces accurate prostate auto-contours for radiotherapy planning. The availability of multiple contours for each CT slice enables clinicians to determine trade-offs in selecting the "best fitting" contour on each CT slice. Mitigation of outlier contours in the training dataset improves prediction accuracy, but one must be wary of reduction in variability in the training dataset.

The most important requirements should be available on any watermarking systems which are the robustness against possible attacks and the quality of the watermarked images. In most applications, the watermarking algorithm embeds the watermark have to be robust against possible attacks and keep the quality of the host media as possible. The relationship between the two requirements is completely conflict. In this study, the method focuses on the robustness against chosen attacks for the watermarked image based on Dual Intermediate Significant Bit (DISB) model. This method requires embedding two bits into every pixel of the original image, while and the other six bits are changed so as to directly assimilate the original pixel. In the case, when the two hidden bits are equal or not equal to the original bits, there is a need to use mathematical equations to solve this problem which derived and applied in this study. The results show that the proposed model produces robustness watermarked images as compared to our previous method when focuses on the high quality of the watermarked image which proved that proposed method is better as compared with the Least Significant Bit (LSB) after embedding two bits. The best values investigated when the Peak Signal to Noise Ratio (PSNR) is equal or more than 30db, and finding the best Normalized Cross Correlation (NCC) to evaluate the image resistance against attacks. The best values investigated when the two embedded bits are k1=4 and k2=6.