In many real applications, traditional super-resolution (SR) methods fail to provide high-resolution images due to objectionable blur and inaccurate registration of input low-resolution images. Only integer resolution enhancement factors, such as 2 or 3, are often considered, but non-integer factors between 1 and 2 are also important in real cases. We introduce a method to SR and deconvolution, which assumes no prior information about the shape of degradation blurs, incorporates registration parameters, and is properly defined for any rational (fractional) resolution factor. The method minimizes a regularized energy function with respect to the high-resolution image and blurs, where regularization is carried out in both the image and blur domains. The blur regularization is based on a generalized multi-channel blind deconvolution constraint derived in the paper. An extension to color images is briefly discussed. Experiments on real data illustrate robustness to noise and other advantages of the method.

This paper presents a new approach to the blind deconvolution and superresolution problem of multiple degraded low-resolution frames of the original scene. We do not assume any prior information about the shape of degradation blurs. The proposed approach consists of building a regularized energy function and minimizing it with respect to the original image and blurs, where regularization is carried out in both the image and blur domains. The image regularization based on variational principles maintains stable performance under severe noise corruption. The blur regularization guarantees consistency of the solution by exploiting differences among the acquired low-resolution images. Several experiments on synthetic and real data illustrate the robustness and utilization of the proposed technique in real applications.

Gabor representations present a number of interesting properties despite the fact that the basis functions are nonorthogonal and provide an overcomplete representation or a nonexact reconstruction. Overcompleteness involves an expansion of the number of coefficients in the transform domain and induces a redundancy that can be further reduced through computational costly iterative algorithms like Matching Pursuit. Here, a biologically plausible algorithm based on competitions between neighboring coefficients is employed for adaptively representing any source image by a selected subset of Gabor functions. This scheme involves a sharper edge localization and a significant reduction of the information redundancy, while, at the same time, the reconstruction quality is preserved. The method is characterized by its biological plausibility and promising results, but it still requires a more in depth theoretical analysis for completing its validation.

We present here a sparse coding algorithm dedicated exclusively to overcomplete Gabor wavelets and based on perceptual contour extraction methods. The algorithm has reduced computational cost and achieves compression rates similar to the standard image compression algorithms (JPEG and JPEG-2000) with the additional advantage of limiting the appearance of artifacts.

We present a single-scale architecture for both segregation and recovering of intrinsic image features and brightness perception. Specifically, a given intensity (or grey scale) image is first analyzed for texture (here defined as small-scale even symmetric features), surfaces (small-scale odd symmetric features) and gradients (large-scale even and odd symmetric features). In this way the image is segregated. Subsequently, textures, surfaces and gradients are recovered by corresponding neural circuits. The proposed architecture may serve as a generic building block for a variety of early vision tasks such as, for example, denoising, efficient coding, as well as mid-level tasks that build on the results from the preceding processing stages.

This paper proposes an efficient method for lossless coding of gray level edge images, which are frequently used in image processing. The method consists of analyzing the edge image in connected sets of pixels by means of chain coding as a function of its starting points, their relative intensities and their displacements within these sets. In order to avoid the redundant information a DPCM coder is used. Then the pixel intensities, the costly information, are coded by an arithmetic encoder, and the rest (coordinates, displacements and movements) by a Huffman encoder, which is less demanding in terms of hardware and software than the first one. For this type of images, the proposed method provides lossless compression rates higher than the commonly used lossless methods. Results show that the proposed method can provide for this type of images 20-78% better compression when compared with the best available methods, and 19% better compression when compared with JPEG-LS (arithmetic version).

The paper evaluates the utility of coherent optical systems with invariant color correlation for the recognition of vibrio cholerae 01 in culturable and nonculturable stage stained with direct immunofluorescence in laboratory and environmental samples. Images of scenes was recorded with CCD camera and decomposed in three RGB channels. The position, scale and rotation invariant image recognition was made through the scale transform. In all cases the bacteria was identified. The correlation peak positions that appear in green channel output are dependent of differences of bacteria's angle (along y-axis) and size (along x-axis) in problem images with the filter image.

We present methods for obtaining basic signal parameters and apply them to the problem of detection, estimation, and enhancement of biological signals. We apply these methods to human speech and whale sounds. In speech, the methods are used to obtain statistical properties of the signal so that the derived features may be used to cluster speakers. For whale sounds we show that the methods may be effectively used not only to obtain signal components and parameters but to also adaptively filter off the noise.

We present a number of autofocusing methods in light microscopy for use in diatom identification. Among these, the Tenengrad method has been considered one of the best. The basic requirements for a practical autofocusing system are speed, sharpness and robustness to noise. Recently other focus measures based on a modified Laplacian method are said to perform better than Tenengrad. We investigate two sound methods based on a modified Tenengrad and a modified Laplacian. Measurements show that they provide a reliable and suitable focus measure that outperform similar methods. We investigate the window size analysis dependency and perform an univariate analysis on the focus measures. The focusing techniques are implemented in an automatic slide scanning system for diatom detection and identification for its use in the ADIAC project.