This PDF file contains the editorial “Editorial” for JEI Vol. 5 Issue 01

A Markov source model is described for a simple subset of printed music notation that was developed as an extended example of the document image decoding (DID) approach to document image analysis. The model is based on the Adobe Sonata music symbol set and a finite-state language of textual music messages. The music message language is defined and several important aspects of message imaging are discussed. Aspects of music notation that appear problematic for a finite-state representation are identified. Finally, an example of music image decoding and resynthesis using the model is presented. Development of the model was greatly facilitated by the duality between image synthesis and image decoding that is fundamental to the DID paradigm.

The Choquet fuzzy integral is applied to handwritten word recognition. A handwritten word recognition system is described. The word recognition system assigns a recognition confidence value to each string in a lexicon of candidate strings. The system uses a lexicon-driven approach that integrates segmentation and recognition via dynamic programming matching. The dynamic programming matcher finds a segmentation of the word image for each string in the lexicon. The traditional match score between a segmentation and a string is an average. In this paper, fuzzy integrals are used instead of an average. Experimental results demonstrate the utility of this approach. A surprising result is obtained that indicates a simple choice of fuzzy integral works better than a more complex choice.

Text structures in document images are usually laid out in a structured manner—having preferred spatial relations. These spatial relations are rarely deterministic; however, they can be modeled by probabilities. Therefore, any realistic document layout analysis algorithm should utilize this type of probabilistic knowledge to optimize its performance. We first describe a method for automatically generating a large amount of nearly perfect layout ground truth data from the LaTeX device-independent (DVI) files, where the bounding boxes for the characters, words, text lines, and text blocks are represented in hierarchies. These ground truth data enable us to construct statistical models that characterize the various layout structures in document images. We demonstrate this concept through the development of a word segmentation algorithm, which employs the recursive morphological closing transform to model word shapes in document images. We also conducted systematic experiments to evaluate the performance of our algorithm using the synthetic images generated from the LaTeX DVI files and the real images from the UW-I and UW-II English document image databases. The results indicate that the correct word detection rate is about 95% on the synthetic images and more than 90% on most of the tested real images.

A system for detecting and locating user-specified search strings, or phrases, in lines of imaged text is described. The phrases may be single words or multiple words, and may contain a partially specified word. The imaged text can be composed of a number of different fonts and graphics. Textlines in a deskewed image are hypothesized using multiresolution morphology. For each textline, the baseline, topline and x-height are identified by simple statistical methods and then used to normalize each textline bounding box. Columns of pixels in the resulting bounding box serve as feature vectors. One hidden Markov model is created for each user-specified phrase and another represents all text and graphics other than the user-specified phrases. Phrases are identified using Viterbi decoding on a spotting network created from the models. The operating point of the system can be varied to trade off the percentage of words correctly spotted and the percentage of false alarms. Results are given using a subset of the UW English Document Image Database I.

Images scanned in the presence of mechanical vibration are subject to artifacts such as brightness fluctuation and geometric warping. The goal of the present study is to characterize these distortions and develop a restoration algorithm to invert them, hence producing an output digital image consistent with a scanner operating under ideal uniform motion conditions. The image restoration algorithm described in this paper makes use of the instantaneous velocity of the linear sensor array to reconstruct an underlying piecewise constant or piece-wise linear model of the image irradiance profile. That reconstructed image is then suitable for resampling under ideal scanning conditions to produce the restored output digital image. We demonstrate the algorithm on simulated scanned imagery with typical operating parameters.

For given binary image and degradation processes, an optimal mean-absolute-error translation-invariant filter can be designed via the representation of such filters as a union of morphological hit-or-miss transforms. The present paper investigates a different optimization methodology by representing translation-invariant filters as differencing filters. Rather than employing structuring templates to build the entire output image, as is done with direct hit-or-miss representation, differencing filters only employ templates that locate value flips (black-to-white or white-to-black). Differencing filters play a central role in several digital document processing tasks and the paper considers their optimal design. The paper compares the logic designs of differencing and direct hit-or-miss representations, the combinational logic costs of the two representations, and the estimation precision of optimization approaches based on each. Both combinational logic cost and precision are relative to image models. It is also shown how differencing filters are statistically designed and applied in the digital document setting for image restoration and resolution conversion.

Accurate color reproduction of images presented on a computer-controlled CRT display as projected 35-mm transparencies is a complicated procedure requiring the characterization and control of several imaging processes and the application of appropriate color appearance modeling to account for the changes in viewing conditions. We review a process for image recorder characterization, projection system characterization, and testing of color appearance models for this application. Accurate image recorder characterization was achieved through a combination of empirical modeling of the exposure and processing system and of a physical model of photographic film. The projection system characterization included specification of the spectral properties of the light source, reflectance properties of the viewing screen, and the effects of light exposure and temperature on the photographic transparencies. Color appearance models were used to predict the changes in image color appearance due to changes in media, white point, luminance, and surround. The RLAB model proved to work best in this application.

Taking colorimetric color reproduction into account, the conventional error diffusion method is modified for color digital half-toning. Assuming that the input to a bilevel color printer is given in CIE-XYZ tristimulus values or CIE-LAB values instead of the more conventional RGB or YMC values, two modified versions based on vector operation in (1) the XYZ color space and (2) the LAB color space were tested. Experimental results show that the modified methods, especially the method using the LAB color space, resulted in better color reproduction performance than the conventional methods. Spatial artifacts that appear in the modified methods are presented and analyzed. It is also shown that the modified method (2) with a thresholding technique achieves a good spatial image quality.

This PDF file contains the editorial “Book Rvw: Foundations of Vision," by Brian A. Wandell” for JEI Vol. 5 Issue 01