Image matching is a critical process in various photogrammetry, computer vision and remote sensing applications such as image registration, 3D model reconstruction, change detection, image fusion, pattern recognition, autonomous navigation, and digital elevation model (DEM) generation and orientation. The primary goal of the image matching process is to establish the correspondence between two images of the same scene (i. e., the reference and input images). Image matching methods are generally classified as feature-based matching and template matching. Feature-based methods extract image features (points, lines, regions) and attempt to establish the correspondence between these features. Template matching methods, also known as area-based methods, are generally defined as the process of finding a template in an image, based on a similarity measure such as cross-correlation and mutual information. Identical image windows of predefined size are applied for the computation of correspondence. Similarity measures play an essential role in the quality of template matching in photogrammetry, remote sensing, and computer vision. Various similarity measures have been proposed in the literature. Each similarity measure has its strengths and weaknesses. In this paper, the capability of some well-known similarity measures for matching of various close range and satellite images with diverse geometric and radiometric differences are evaluated. Also, to increase the template matching stability against geometric and radiometric variations, a novel weighting approach for computing of similarity measures has been introduced. The proposed approach is based on three weight factor that are computed using gradient and Gaussian functions. By applying this weighting approach for cross correlation similarity measure, a novel measure named Weighted Cross-Correlation (WCC) has been presented. Ten algorithms, including SSD (Sum of Squared Differences), LSSSD (Locally Scaled Sum of Squared Differences), NSSD (Normalized Sum of Squared Differences), JF (Jeffrey Divergence), Tanimoto, ISD (Incremental Sign Distance), IRV (Intensity-Ratio Variance), CC (Cross-Correlation), MI (Mutual Information) and WCC are considered for evaluation. To evaluate the capability of various similarity measures, a number of template-matching experiments were applied. Several synthetic and real images for different geometric and radiometric variations including, scale, rotation, viewpoint, blur, and illumination changes are used as data set. The similarity measures are evaluated using three evaluation criteria, including success rate, positional accuracy, and computation time. The experimental results indicate that the proposed WCC method outperforms the other similarity measures for all images and all types of transformations. Based on the evaluation results, the WCC method can be applied to the reliable template matching for a variety of photogrammetric and remote sensing applications. Generally, after the WCC, better results, on average, were obtained by the NSSD, LSSSD, and CC measures in most cases. For illumination variations, MI and ISD methods provide the best results. The fastest method is the IRV and the slowest method is MI. Evaluation of the performance of the various similarity measures for other applications such as dense matching process is suggested as future work.

Due to the rapid growth of digital libraries, digitizing large documents has become an important topic. In a quite long book, similar characters, sub-words and words will occur many times. In this paper, we propose a sub-word image clustering method for the applications dealing with large uniform documents. We assumed that the whole document is printed in a single font and print quality is not good. To test our method, we created a dataset of all sub-words of a Farsi book. The book has 233 pages with more than 111000 sub-words manually labeled. We use an incremental clustering algorithm. Four simple features are extracted from each sub-word and compared with the corresponding features of each cluster center. If all features' differences lie within certain thresholds, the sub-word and the winner cluster center are finely compared using a template matching algorithm. In our experiments, we show that all sub-words of the book are recognized with more than 99.7% accuracy by assigning the label of each cluster center to all of its members.

Computer aided pulmonary nodule detection has been among major research topics lately to help the early treatment of lung cancer which is the most lethal kind of cancer worldwide. Some evidence suggests that periodic screening tests with the CT of patients will help in reducing the mortality rate caused by the lung cancer. A complete and accurate computer aided diagnosis (CAD) system for detection of nodules in lung CT images consists of three main steps: extraction of lung parenchyma, candidate nodule detection and false positive reduction. While precise segmentation of lung region speed upthe detection process of pulmonary nodules by limiting the search area, in candidate nodule detection step we attempt to include all nodule like structures. However, the main problem in the current CAD systems for nodule detection is the high false positive rate which is mostly associated to misrecognition of juxta-vascular nodules from blood vessels. In this paper we propose an automated method which has all of the three above mentioned steps. Our method attempts to find initial nodules by thresholding and template matching. To separate false positives from nodules, we use feature extraction and neural classifier. The proposed method has been evaluated against several images in LIDC database and the results demonstrate improvements in comparison with the previous methods.

Template matching is a widely used technique in many of image processing and machine vision applications. In this paper we propose a new as well as a fast and reliable template matching algorithm which is invariant to Rotation, Scale, Translation and Brightness (RSTB) changes. For this purpose, we adopt the idea of ring projection transform (RPT) of image. In the proposed algorithm, two novel suggestions are offered that significantly increase the precision and performance of the previous methods. First, our algorithm works with Log-Spectrum of image instead of the image itself, this change increases the accuracy of matching, and secondly for boosting the speed of the searching strategy, a new and modified version of Imperialist Competitive Algorithm, MICA, is presented. This matching procedure avoids the searching algorithm from being trapped in local minimum by taking advantage of adding a modification step to ICA. The simulation results show the superiority of proposed method in comparison with the previous ones.

The traditional method for studying non-stationary signals is spectrogram based on the short-time Fourier transform (STFT). The well known limitation of the STFT is the inherent trade-off between time and frequency resolution. The Wigner-Ville (WV) distribution has the best time-frequency resolution, but its draw back is generating cross-terms. The matching pursuit (MP) distribution based on using the Gaussian atom is always positive, does not include crossterm, and has convenient resolution. In this paper, we have shown in addition to the known properties, the MP distribution can also remove the additive noise inherently. On the other words, we are able to remove the noise just by limiting the algorithm iterations and without paying any additional cost. Although the MP distribution based on using the Gaussian atoms is always positive and it has convenient resolution, according to the MP the time marginal and the frequency marginal will not be obtained accurately. In this paper, it has been shown that by implementing the minimum cross entropy (MCE) technique according to the MP distribution as a priory positive distribution, the new extracted distribution has the most similarity to the MP distribution and it also satisfies the correct time and frequency marginal.

Ground- Penetrating Radar (GPR) is a non-destructive and high-resolution geophysical method that uses high frequency reflected EM waves to detect buried objects and manmade structures. In current study this method has been used to identify geometrical characteristics of buried cylindrical targets such as tunnel structures. This aim has been obtained through determination of relationships between physical and geometrical characteristics of cylindrical targets with the parameters of GPR hyperbolic response using two intelligent pattern recognition methods known as artificial neural network and template matching. To this goal GPR responses of synthetic cylindrical objects corresponding to common geotechnical targets (such as tunnels, canals, qanats and pipes) have been simulated using forward modeling by 2D Finite Difference and have been used as templates in the neural network and template matching algorithms. The structure of applied neural network was designed based on extracting discriminant and unique features (eigen values and the norm of eigen values in the horizontal and vertical directions) from the GPR images and predicting all geometrical parameters of the objects simultaneously. The template matching operation also carried out by two different similarity approaches named spatial domain convolution and normalized cross correlation in 2D wave number domain. Afterward it was delineated that the wave number domain approach is generally faster (more than 23 times) than the other approach.The results of the research show that both two employed intelligent methods having in situ, real-time, accurate and automatic application capabilities can be applied for real geotechnical applications, however in general the neural network method has led to less error and as a result higher estimation power for the geometrical parameters of the cylindrical targets than template matching method.