Stereo Image Matching Using Adaptive Morphological Correlation.

A stereo matching method based on adaptive morphological correlation is presented. The point correspondences of an input pair of stereo images are determined by matching locally adaptive image windows using the suggested morphological correlation that is optimal with respect to an introduced binary dissimilarity-to-matching ratio criterion. The proposed method is capable of determining the point correspondences in homogeneous image regions and at the edges of scene objects of input stereo images with high accuracy. Furthermore, unknown correspondences of occluded and not matched points in the scene can be successfully recovered using a simple proposed post-processing. The performance of the proposed method is exhaustively tested for stereo matching in terms of objective measures using known database images. In addition, the obtained results are discussed and compared with those of two similar state-of-the-art methods.

Deep-learning based detection of COVID-19 using lung ultrasound imagery.

BACKGROUND: The COVID-19 pandemic has exposed the vulnerability of healthcare services worldwide, especially in underdeveloped countries. There is a clear need to develop novel computer-assisted diagnosis tools to provide rapid and cost-effective screening in places where massive traditional testing is not feasible. Lung ultrasound is a portable, easy to disinfect, low cost and non-invasive tool that can be used to identify lung diseases. Computer-assisted analysis of lung ultrasound imagery is a relatively recent approach that has shown great potential for diagnosing pulmonary conditions, being a viable alternative for screening and diagnosing COVID-19. OBJECTIVE: To evaluate and compare the performance of deep-learning techniques for detecting COVID-19 infections from lung ultrasound imagery. METHODS: We adapted different pre-trained deep learning architectures, including VGG19, InceptionV3, Xception, and ResNet50. We used the publicly available POCUS dataset comprising 3326 lung ultrasound frames of healthy, COVID-19, and pneumonia patients for training and fine-tuning. We conducted two experiments considering three classes (COVID-19, pneumonia, and healthy) and two classes (COVID-19 versus pneumonia and COVID-19 versus non-COVID-19) of predictive models. The obtained results were also compared with the POCOVID-net model. For performance evaluation, we calculated per-class classification metrics (Precision, Recall, and F1-score) and overall metrics (Accuracy, Balanced Accuracy, and Area Under the Receiver Operating Characteristic Curve). Lastly, we performed a statistical analysis of performance results using ANOVA and Friedman tests followed by post-hoc analysis using the Wilcoxon signed-rank test with the Holm's step-down correction. RESULTS: InceptionV3 network achieved the best average accuracy (89.1%), balanced accuracy (89.3%), and area under the receiver operating curve (97.1%) for COVID-19 detection from bacterial pneumonia and healthy lung ultrasound data. The ANOVA and Friedman tests found statistically significant performance differences between models for accuracy, balanced accuracy and area under the receiver operating curve. Post-hoc analysis showed statistically significant differences between the performance obtained with the InceptionV3-based model and POCOVID-net, VGG19-, and ResNet50-based models. No statistically significant differences were found in the performance obtained with InceptionV3- and Xception-based models. CONCLUSIONS: Deep learning techniques for computer-assisted analysis of lung ultrasound imagery provide a promising avenue for COVID-19 screening and diagnosis. Particularly, we found that the InceptionV3 network provides the most promising predictive results from all AI-based techniques evaluated in this work. InceptionV3- and Xception-based models can be used to further develop a viable computer-assisted screening tool for COVID-19 based on ultrasound imagery.

Fast Recursive Computation of Sliding DHT with Arbitrary Step.

Short-time (sliding) transform based on discrete Hartley transform (DHT) is often used to estimate the power spectrum of a quasi-stationary process such as speech, audio, radar, communication, and biomedical signals. Sliding transform calculates the transform coefficients of the signal in a fixed-size moving window. In order to speed up the spectral analysis of signals with slowly changing spectra, the window can slide along the signal with a step of more than one. A fast algorithm for computing the discrete Hartley transform in windows that are equidistant from each other is proposed. The algorithm is based on a second-order recursive relation between subsequent equidistant local transform spectra. The performance of the proposed algorithm with respect to computational complexity is compared with the performance of known fast Hartley transform and sliding algorithms.

Target recognition under nonuniform illumination conditions.

A two-step algorithm for reliable recognition of a target imbedded into a two-dimensional nonuniformly illuminated and noisy scene is presented. The input scene is preprocessed with a space-domain pointwise procedure followed by an optimum correlation. The preprocessing is based on an estimate of the source illumination function, whereas the correlation filter is optimized with respect to the mean-squared-error criterion for detecting a target in the preprocessed scene. Computer simulations are provided and compared with those of various common techniques in terms of recognition performance and tolerance to nonuniform illumination as well as to additive noise. Experimental optodigital results are also provided and discussed.

Design of correlation filters for recognition of linearly distorted objects in linearly degraded scenes.

Generalized correlation filters are proposed to improve recognition of a linearly distorted object embedded in a nonoverlapping background when the input scene is degraded with a linear system and additive noise. Several performance criteria defined for the nonoverlapping signal model are used for the design of filters. The derived filters take into account information about an object to be recognized, disjoint background, noise, and linear degradations of the target and the input scene. Computer simulation results obtained with the proposed filters are discussed and compared with those of various correlation filters in terms of discrimination capability, location errors, and tolerance to input noise.

Adaptive phase-input joint transform correlator.

An adaptive phase-input joint transform correlator for pattern recognition is presented. The input of the system is two phase-only images: input scene and reference. The reference image is generated with a new iterative algorithm using phase-only synthetic discriminant functions. The algorithm takes into account calibration lookup tables of all optoelectronics elements used in optodigital experiments. The designed adaptive phase-input joint transform correlator is able to reliably detect a target and its distorted versions embedded into a cluttered background. Computer simulations are provided and compared with those of various existing joint transform correlators in terms of discrimination capability, tolerance to input additive noise, and to small geometric image distortions. Experimental optodigital results are also provided and discussed.

Pattern recognition with an adaptive joint transform correlator.

An adaptive joint transform correlator for real-time pattern recognition is presented. A reference image for the correlator is generated with a new iterative algorithm. The training algorithm is based on synthetic discriminant functions. The obtained reference image contains the information needed to reliably discriminate a target against known false objects and a cluttered background. Calibration lookup tables of all optoelectronics elements used are included in the design of the adaptive joint transform correlator. Two methods for the implementation of the proposed joint transform correlator in an optodigital setup are considered. Experimental results are provided and compared with those of computer simulations.