Correlation-based Template Matching Research Articles

Extended framework for Sindhi numerals OCR using gradient orientation histograms

The accuracy on MINST dataset for roman numerals is already 99.65%. However, same models showed low accuracy on Sindhi numerals. It is because Sindhi numerals have high correlation between the shapes of the numerals. In this paper, correlation based template matching is used to analyze the shape ambiguity by identifying the dominant false positives (FP) and false negatives (FN) for every numeral. Furthermore, the Gradients Histogram Orientation (GOH) features are used to improve the accuracy of existing classifiers by image-to-image matching. The classical OCR using simple binary features are not sufficient to address the problems of shape ambiguity in Sindhi numerals, i.e., the shape of digits 2, , and 3, , are very similar. The raw pixel values are used as features for the classification in the first stage. In second stage, the input image is matched with the dominant FP and FN of the predicted class, and the final decision is made by the image-to-image matching based on GOH features. Decision based on image to image matching with dominant FP and FN increase the accuracy of the classifier. Support vector machine (SVM), K-nearest neighbor, and template based matching classifiers are used. The proposed extension substantially improves the accuracy of all mentioned classifiers.

Recognizing Solo Jazz Dance Moves Using a Single Leg-Attached Inertial Wearable Device

We present here a method for recognising dance moves in sequences using 3D accelerometer and gyroscope signals, acquired by a single wearable device, attached to the dancer’s leg. The recognition entails dance tempo estimation, temporal scaling, a wearable device orientation-invariant coordinate system transformation, and, finally, sliding correlation-based template matching. The recognition is independent of the orientation of the wearable device and the tempo of dancing, which promotes the usability of the method in a wide range of everyday application scenarios. For experimental validation, we considered the versatile repertoire of solo jazz dance moves. We created a database of 15 authentic solo jazz template moves using the performances of a professional dancer dancing at 120 bpm. We analysed 36 new dance sequences, performed by the professional and five recreational dancers, following six dance tempos, ranging from 120 bpm to 220 bpm with 20 bpm increment steps. The recognition F1 scores, obtained cumulatively for all moves for different tempos, ranged from 0.87 to 0.98. The results indicate that the presented method can be used to recognise repeated dance moves and to assess the dancer’s consistency in performance. In addition, the results confirm the potential of using the presented method to recognise imitated dance moves, supporting the learning process.

An accurate and distraction-free vision-based structural displacement measurement method integrating Siamese network based tracker and correlation-based template matching

Vision-based displacement measurement receives increasing attention on non-contact bridge monitoring while it faces challenges in long-time field applications due to the presence of environmental variations. To overcome this issue, this study proposes a novel distraction-free displacement measurement approach by integrating deep learning-based Siamese tracker with correlation-based template matching. The Siamese tracker used applies deep feature representations and learned similarity measures for image matching and also considers adaptive template update with time. Since the estimated bounding boxes by the Siamese tracker have size changes within frame sequences, a correction step is added to remove the centroid drifts between the template and the predicted target regions using correlation-based template matching. The proposed method is validated first in an indoor test and then implemented in monitoring tests on a short-span footbridge and a long-span road bridge, demonstrating its potential to handle challenging scenarios including partial occlusion, illumination changes, background variations and shade effects.

MRI artifact simulation for clinically relevant MRI sequences for guidance of prostate HDR brachytherapy

For the purpose of magnetic resonance imaging (MRI) guidance of prostate high-dose-rate (HDR) brachytherapy, this paper presents a study on the potential of clinically relevant MRI sequences to facilitate tracking or localization of brachytherapy devices (HDR source/titanium needle), and which could simultaneously be used to visualize the anatomy. The tracking or localization involves simulation of the MRI artifact in combination with a template matching algorithm. Simulations of the MRI artifacts induced by an HDR brachytherapy source and a titanium needle were implemented for four types of sequences: spoiled gradient echo, spin echo, balanced steady-state free precession (bSSFP) and bSSFP with spectral attenuated inversion recovery (SPAIR) fat suppression. A phantom study was conducted in which mentioned sequences (in 2D) as well as the volumetric MRI sequences of the current clinical scan protocol were applied to obtain the induced MRI artifacts for an HDR source and a titanium needle. Localization of the objects was performed by a phase correlation based template matching algorithm. The simulated images demonstrated high correspondences with the acquired MR images, and allowed localization of the objects. A comparison between the object positions obtained for all applied MRI sequences showed deviations (from the average position) of 0.2–0.3 mm, proving that all MRI sequences were suitable for localization of the objects, irrespective of their 2D or volumetric nature. This study demonstrated that the MRI artifact induced by an HDR source or a titanium needle could be simulated for the four investigated types of MRI sequences (spoiled gradient echo, spin echo, bSSFP and bSSFP-SPAIR), valuable for real-time object localization in clinical practice. This leads to more flexibility in the choice of MRI sequences for guidance of HDR brachytherapy, as they are suitable for both object localization and anatomy visualization.

Vision-based systems for structural deformation measurement: case studies

Vision-based systems offer a promising way to measure displacement and are receiving increased attention in civil structural monitoring. However, the working performance of vision-based systems, especially their measurement accuracy and robustness to different field conditions, is not fully understood. This study reports three cases studies of vision-based monitoring tests including one in a laboratory, one on a short-span bridge and one on a long-span bridge. The tracking accuracy is quantified in laboratory conditions in the range from 0·02 to 0·20 pixel, depending on the target patterns as well as the tracking method selected. The measurement performance under several field challenges is investigated, including long-range measurement (e.g. camera-to-target distance of 710 m), low-contrast target patterns, changes of target patterns and changes in lighting conditions. Three representative tracking methods for the video processing, namely, correlation-based template matching, Lucas–Kanade optical flow estimation and scale-invariant feature transform were used for the analysis, indicating their advantages and shortcomings for field measurement. One of the main observations in field application is that changes in lighting conditions might cause some low-frequency measurement errors that could be misunderstood without prior knowledge regarding structural loading conditions.

Reconfigurable Instruction-Based Multicore Parallel Convolution and Its Application in Real-Time Template Matching

Convolution is widely used in scientific computational fields such as digital image processing and machine learning. However, these applications are difficult to execute in realtime because they are computationally intensive. This paper introduces a high-speed convolution solution that runs on our self-developed multicore digital signal processor (DSP). To optimize the convolution capability, we propose a convolution instruction and a convolution micro architecture in the design of a subcore. As a coprocessor, the designed subcore is integrated into a network-on-chip (NoC)-based multicore DSP. In the implementation of the multicore parallel convolution, an independent convolution task-partitioning and mapping scheme is proposed. Datablock storage and software prefetching mechanisms are used to hide the data transmission time during the calculation, improving the computing efficiency. We also develop a data reuse strategy that effectively reduces the data bandwidth requirements of multicore parallel convolution. The proposed methods are applied to correlation-based template matching, with the results showing that our convolution computing approach greatly improves the performance compared with the same operations run on a personal computer, a TMS320C6678 processor and an NVIDIA Quadro 1000M graphics processing unit (GPU).

Reliable Subpixel Ground Control Point Estimation Algorithm Using Vector Roads

This letter proposes a novel algorithm for reliable and accurate ground control point (GCP) estimation, which uses vector road network data as a reference. Roads are detected in satellite images, and road patches are used as a template, which is matched against the vector road data by using a nonlinear criterion based on the distance transform. The proposed approach is compared with the standard correlation-based template matching in terms of the reliability and accuracy of localization on a test set of five RapidEye images with 380 reference control points. The experiments demonstrate superior reliability with respect to the size of the search window and the variable image acquisition conditions. While the subpixel accuracy is achieved by the cross-correlation method over the search area up to 20% of the template size, the proposed method achieves comparable accuracy over a search area that is ten times larger than the template image.

Respiratory motion tracking using Microsoft’s Kinect v2 camera

Abstract In image-guided radiotherapy, monitoring and compensating for respiratory motion is of high importance. We have analysed the possibility to use Microsoft’s Kinect v2 sensor as a low-cost tracking camera. In our experiment, eleven circular markers were printed onto a Lycra shirt and were tracked in the camera’s color image using cross correlation-based template matching. The 3D position of the marker was determined using this information and the mean distance of all template pixels from the sensor. In an experiment with four volunteers (male and female) we could demonstrate that real time position tracking is possible in 3D. By averaging over the depth values inside the template, it was possible to increase the Kinect’s depth resolution from 1 mm to 0.1 mm. The noise level was reduced to a standard deviation of 0.4 mm. Temperature sensitivity of the measured depth values was observed for about 10-15 minutes after system start.

Detection of reappearing targets in forward-looking infrared video sequences

Often sensor ego-motion or fast target movement causes the target to temporarily go out of the field-of-view leading to reappearing target detection problem in target tracking applications. Since the target goes out of the current frame and re-enters at a later frame, the re-entering location and variations in rotation, scale, and other three-dimensional orientations of the target are not known, thus complicating the detection and tracking of reappearing targets. A new training-based target detection algorithm has been developed using tuned basis functions (TBFs). The detection algorithm uses target and background information, extracted from training samples, to detect possible candidate target images. The detected candidate target images are then introduced into the second algorithm, called clutter rejection module, to determine the target re-entering frame and location of the target. The second algorithm has been designed using the spatial domain correlation-based template matching (TM) technique. If the target re-enters the current frame, the target coordinates are detected and tracking algorithm is initiated. The performance of the proposed TBF-TM-based reappearing target detection algorithm has been tested using real-world forward-looking infrared video sequences.

Hybrid markerless tracking of complex articulated motion in golf swings

Sports video tracking is a research topic that has attained increasing attention due to its high commercial potential. A number of sports, including tennis, soccer, gymnastics, running, golf, badminton and cricket have been utilised to display the novel ideas in sports motion tracking. The main challenge associated with this research concerns the extraction of a highly complex articulated motion from a video scene. Our research focuses on the development of a markerless human motion tracking system that tracks the major body parts of an athlete straight from a sports broadcast video. We proposed a hybrid tracking method, which consists of a combination of three algorithms (pyramidal Lucas-Kanade optical flow (LK), normalised correlation-based template matching and background subtraction), to track the golfer's head, body, hands, shoulders, knees and feet during a full swing. We then match, track and map the results onto a 2D articulated human stick model to represent the pose of the golfer over time. Our work was tested using two video broadcasts of a golfer, and we obtained satisfactory results. The current outcomes of this research can play an important role in enhancing the performance of a golfer, provide vital information to sports medicine practitioners by providing technically sound guidance on movements and should assist to diminish the risk of golfing injuries.

Automatic recognition method for optical measuring instruments based on machine vision

Based on a comprehensive study of various algorithms, the automatic recognition of traditional ocular optical measuring instruments is realized. Taking a universal tools microscope(UTM) lens view image as an example, a 2-layer automatic recognition model for data reading is established after adopting a series of pre-processing algorithms. This model is an optimal combination of the correlation-based template matching method and a concurrent back propagation(BP) neural network. Multiple complementary feature extraction is used in generating the eigenvectors of the concurrent network. In order to improve fault-tolerance capacity, rotation invariant features based on Zernike moments are extracted from digit characters and a 4-dimensional group of the outline features is also obtained. Moreover, the operating time and reading accuracy can be adjusted dynamically by setting the threshold value. The experimental result indicates that the newly developed algorithm has optimal recognition precision and working speed. The average reading ratio can achieve 97.23%. The recognition method can automatically obtain the results of optical measuring instruments rapidly and stably without modifying their original structure, which meets the application requirements.

Anaglyph stereo visualization by the use of a single image and depth information

In this paper, an anaglyph stereo visualization is studied by the use of a single image and depth information. The present technique allows the stereo visualization of the target image without binocular camera, which has been used for generating the standard anaglyphs. Three test cases are shown in this work, which covers the generation of monochrome anaglyph from given geometrical information, and that of monochrome and color anaglyph from depth information evaluated from other imaging devises, which are placed in parallel. The experimental depth information of the target image is evaluated from the correlation-based template matching analysis of the random dot patterns optically projected on the target objects. Examples of anaglyph stereo visualization are shown for scientific art, such as the monochrome cube, monochrome plaster figure and the color artificial flowers. These results indicate that the present technique of anaglyph stereo visualization through the depth information is very useful and widely applicable to the general three-dimensional visualization.