Jeff Orchard

A nonlocal-means approach to exemplar-based inpainting

This paper introduces a novel approach to the problem of image inpainting through the use of nonlocal-means. In traditional inpainting techniques, only local information around the target regions are used to fill in the missing information, which is insufficient in many cases. More recent inpainting techniques based on the concept of exemplar-based synthesis utilize nonlocal information but in a very limited way. In the proposed algorithm, we use nonlocal image information from multiple samples within the image. The contribution of each sample to the reconstruction of a target pixel is determined using an weighted similarity function and aggregated to form the missing information. Experimental results show that the proposed method yields quantitative and qualitative improvements compared to the current exemplar-based approach. The proposed approach can also be integrated into existing exemplar-based inpainting techniques to provide improved visual quality.

Efficient nonlocal-means denoising using the SVD

Nonlocal-means (NL-means) is an image denoising method that replaces each pixel by a weighted average of all the pixels in the image. Unfortunately, the method requires the computation of the weighting terms for all possible pairs of pixels, making it computationally expensive. Some short-cuts assign a weight of zero to any pixel pairs whose neighbourhood averages are too dissimilar. In this paper, we propose an alternative strategy that uses the SVD to more efficiently eliminate pixel pairs that are dissimilar. Experiments comparing this method against other NL-means speed-up strategies show that its refined discrimination between similar and dissimilar pixel neighbourhoods significantly improves the denoising effect.

Efficient Least Squares Multimodal Registration With a Globally Exhaustive Alignment Search

There are many image registration situations in which the initial misalignment of the two images is large. These registration problems, often involving comparison of the two images only within a region of interest (ROI), are difficult to solve. Most intensity-based registration methods perform local optimization of their cost function and often miss the global optimum when the initial misregistration is large. The registration of multimodal images makes the problem even more difficult since it limits the choice of available cost functions. We have developed an efficient method, capable of multimodal rigid-body registration within an ROI, that performs an exhaustive search over all integer translations, and a local search over rotations. The method uses the fast Fourier transform to efficiently compute the sum of squared differences cost function for all possible integer pixel shifts, and for each shift models the relationship between the intensities of the two images using linear regression. Test cases involving medical imaging, remote sensing and forensic science applications show that the method consistently brings the two images into close registration so that a local optimization method should have no trouble fine-tuning the solution.

Cut-out image mosaics

An image mosaic is a rendering of a large target image by arranging a collection of small source images, often in an array, each chosen specifically to fit a particular block of the target image. Most mosaicking methods are simplistic in the sense that they break the target image into regular tiles (e.g., squares or hexagons) and take extreme shortcuts when evaluating the similarity between target tiles and source images. In this paper, we propose an efficient method to obtain higher quality mosaics that incorporate a number of process improvements. The Fast Fourier Transform (FFT) is used to compute a more fine-grained image similarity metric, allowing for optimal colour correction and arbitrarily shaped target tiles. In addition, the framework can find the optimal sub-image within a source image, further improving the quality of the matching. The similarity scores generated by these high-order cost computations are fed into a matching algorithm to find the globally-optimal assignment of source images to target tiles. Experiments show that each improvement, by itself, yields a more accurate mosaic. Combined, the innovations produce very high quality image mosaics, even with only a few hundred source images.

Efficient FFT-Accelerated Approach to Invariant Optical–LIDAR Registration

This paper presents a fast Fourier transform (FFT)-accelerated approach designed to handle many of the difficulties associated with the registration of optical and light detection and ranging (LIDAR) images. The proposed algorithm utilizes an exhaustive region correspondence search technique to determine the correspondence between regions of interest from the optical image with the LIDAR image over all translations for various rotations. The computational cost associated with exhaustive search is greatly reduced by exploiting the FFT. The substantial differences in intensity mappings between optical and LIDAR images are addressed through local feature mapping transformation optimization. Geometric distortions in the underlying images are dealt with through a geometric transformation estimation process that handles various transformations such as translation, rotation, scaling, shear, and perspective transformations. To account for mismatches caused by factors such as severe contrast differences, the proposed algorithm attempts to prune such outliers using the random sample consensus technique to improve registration accuracy. The proposed algorithm has been tested using various optical and LIDAR images and evaluated based on its registration accuracy. The results indicate that the proposed algorithm is suitable for the multimodal invariant registration of optical and LIDAR images.

Particle swarm optimization using dynamic tournament topology

Graphical abstractDisplay Omitted HighlightsThis paper proposes a dynamic tournament topology strategy to improve PSO (DTT-PSO).Instead of the gbest, lbest, or other bests, the DTT-PSO chooses several relatively better guides from the entire population for each particle.The strategy ensures that the best particle informs the others with high probability only and not those with necessity.The method also incorporates merits from random topology and fully informed strategy.The particle simultaneously receives information from all better guides. Furthermore, particle guides frequently change with evolution.The method increases population diversity and decreases the probability of dropping into local optima. Furthermore, this paper shows the application of DTT-PSO in optimization of artificial neural networks. Particle swarm optimization (PSO) is a nature-inspired global optimization method that uses interaction between particles to find the optimal solution in a complex search space. The swarms evolving solution is represented by the best solution found by any particle. However, using this best solution often limits the search area. In this paper, we propose a dynamic tournament topology strategy to improve PSO. In our method, each particle is guided by several better solutions, chosen from the entire population. The selection of the better particles is stochastic, but still favors particles with better solutions. Experimental results on benchmark functions indicate that the proposed method is promising. Furthermore, the application of our dynamic tournament topology strategy in optimization of artificial neural networks indicates that this method has favorable performance.

Robust Multimodal Registration Using Local Phase-Coherence Representations

Automatic registration of multimodal images has proven to be a difficult task. Most existing techniques have difficulty dealing with situations involving highly non-homogeneous image contrast and a small initial overlapping region between the images. This paper presents a robust multi-resolution method for regis tering multimodal images using local phase-coherence representations. The proposed method finds the transformation that minimizes the error residual between the local phase-coherence representations of the two multimodal images. The error residual can be minimized using a combination of efficient globally exhaustive optimization techniques and subpixel-level local optimization techniques to further improve robustness in situations with small initial overlap. The proposed method has been tested on various medical images acquired using different modalities and evaluated based on its registration accuracy. The results show that the proposed method is capable of achieving better accuracy than existing multimodal registration techniques when handling situations where image non-homogeneity and small overlapping regions exist.

Fast Discrete Orthonormal Stockwell Transform

We present an efficient method for computing the discrete orthonormal Stockwell transform (DOST). The Stockwell transform (ST) is a time-frequency decomposition transform that is showing great promise in various applications, but is limited because its computation is infeasible for most applications. The DOST is a nonredundant version of the ST, solving many of the memory and computational issues. However, computing the DOST of a signal of length

Improving Neural-Network Classifiers Using Nearest Neighbor Partitioning

N

Registering a MultiSensor Ensemble of Images

using basis vectors is still