Jeffery R. Price

Biased reconstruction for JPEG decoding

Assuming a Laplacian distribution, there exists a well known method for optimally biasing the reconstruction levels for the quantized ac discrete cosine transform (DCT) coefficients in the JPEG decoder. This, however, requires an estimate of the Laplacian distribution parameter. We derive a new, maximum likelihood estimate of the Laplacian parameter using only the quantized coefficients available at the decoder. We quantify the benefits of biased reconstruction through extensive simulations and demonstrate that such improvements are very close to the best possible resulting from centroid reconstruction.

Resampling and reconstruction with fractal interpolation functions

An alternative form of the fractal interpolation function (FIF)-previously unmentioned in the signal processing literature-is noted. This form highlights a simple relationship between fractal and linear interpolation. Using this relationship, many FIF problems can be reduced to a matrix/vector expression. This expression provides a more powerful way to employ the FIF for interpolation and permits its adaptation for reconstruction. Additionally, the alternate form of the FIF allows the construction of fractal functions whose piecewise integrals match observed data.

Adaptive regularized image interpolation using data fusion and steerable constraints

This paper presents an adaptive regularized image interpolation algorithm from blurred and noisy low resolution image sequence, which is developed in a general framework based on data fusion. This framework can preserve the high frequency components along the edge orientation in a restored high resolution image frame. This multiframe image interpolation algorithm is composed of two levels of fusion algorithm. One is to obtain enhanced low resolution images as an input data of the adaptive regularized image interpolation based on data fusion. The other one is to construct the adaptive fusion algorithm based on regularized image interpolation using steerable orientation analysis. In order to apply the regularization approach to the interpolation procedure, we first present an observation model of low resolution video formation system. Based on the observation model, we can have an interpolated image which minimizes both residual between the high resolution and the interpolated images with a prior constraints. In addition, by combining spatially adaptive constraints, directional high frequency components are preserved with efficiently suppressed noise. In the experimental results, interpolated images using the conventional algorithms are shown to compare the conventional algorithms with the proposed adaptive fusion based algorithm. Experimental results show that the proposed algorithm has the advantage of preserving directional high frequency components and suppressing undesirable artifacts such as noise.

Dense range image smoothing using adaptive regularization

We propose an adaptive regularization algorithm for smoothing dense range images using a novel, first order stabilizing function. The stabilizer we suggest is based upon minimizing the reconstructed surface area and is derived in the native, spherical coordinate system of the range scanner. This allows adjustments to be made along only the direction of measurement, thereby preventing the data overlapping problem that can arise in dense images. Adaptation is achieved by adjusting the regularization parameter according to the results of 2D edge analysis. Results indicate effective noise suppression along with well preserved edges and details in the reconstructed, 3D surfaces.

Fractal interpolation of images and volumes

We present a method for constructing fractal interpolation surfaces and volumes through points sampled on rectangular lattices. Unlike other surface constructions ours uses rectangular rather than triangular tilings, halving the number of required parameters. This method is no more complex than previous constructions and yet does not suffer from their limitations. Additionally, our construction extends easily to volumetric interpolation, for which there were no previous (continuous) constructions. In addition to an example with synthetic data, a real image is interpolated using a fractal surface. Limitations and possible improvements are mentioned.

Optimal prefiltering for improved image interpolation

In this paper we derive an optimal (MMSE) prefilter for image interpolation. This derivation is based upon a model of the sensor used to capture the image. To employ this model, we restate the interpolation problem in an intuitive, reconstruction-like fashion. Using a simple CCD sensor model, an example prefilter is derived. Simulations with this prefilter are performed using linear and cubic interpolation as well as an ad hoc, directional interpolation scheme. Quantitative and subjective results indicate that prefiltering generally improves the quality of the interpolated images.

Dequantization bias for JPEG decompression

Standard JPEG decompression reconstructs quantized DCT coefficients to the center for the quantization bin. This fails to exploit the nonuniform distribution of the AC coefficients. Assuming a Laplacian distribution, we derive a maximum likelihood estimate of the Laplacian parameter, based on the quantized coefficients available at the decoder, and use this estimate to optimally bias the reconstruction levels during decompression. As a decoder enhancement, this technique is fully compatible with the JPEG standard and does not modify the JPEG compressed bit stream. Extensive simulations indicate that, at typical compression ratios, biased reconstruction results in modest PSNR improvements-about 0.25 dB or higher-and slight subjective improvements, for little or no computational cost. Furthermore, simulations show that the PSNR improvements are very close (within 0.07 dB) to the best theoretically possible.

Superquadrics-based object representation of complex scenes from range images.

This paper investigates the superquadrics-based object representation of complex scenes from range images. The issues on how the recover-and-select algorithm is incorporated to handle complex scenes containing background and multiple occluded objects are addressed respectively. For images containing backgrounds, the raw image is first coarsely segmented using the scan-line grouping technique. An area threshold is then taken to remove the backgrounds while keeping all the objects. After this pre-segmentation, the recover-and-select algorithm is applied to recover superquadric (SQ) models. For images containing multiple occluded objects, a circle-view strategy is taken to recover complete SQ models from range images in multiple views. First, a view path is planned as a circle around the objects, on which images are taken approximately every 45 degrees. Next, SQ models are recovered from each single-view range image. Finally, the SQ models from multiple views are registered and integrated. These approaches are tested on synthetic range images. Experimental results show that accurate and complete SQ models are recovered from complex scenes using our strategies. Moreover, the approach handling background problems is insensitive to the pre-segmentation error.

Steerable filter cascades

In this paper, we present the notion of cascading steerable filters to improve their angular resolution. Additionally, we illustrate that the results of such cascades can be steered themselves. An advantage of this approach is that only a single, relatively small set of steerable filters can be employed to achieve various angular resolutions. Improving angular resolution has previously required an entirely different, larger set of filters.

Sensor optimal image interpolation

Previously, we derived sensor optimal prefilters for image interpolation. The prefilters were applied prior to integer interpolation with a standard (e.g., linear or cubic) kernel. Here we expand upon that notion and construct complete, sensor optimal interpolation kernels for rational interpolation factors. After restating the interpolation problem in a reconstruction-like fashion, we employ a simple model of the image capture system to derive the MMSE interpolator. Results indicate significant subjective improvements over cubic interpolation, for little extra computation.