Taras Holotyak

Information-theoretic analysis of steganalysis in real images

In this paper we consider the problem of performance improvement of non-blind statistical steganalysis of additive steganography in real images. The proposed approach differs from the existing solutions in two main aspects:(a) a locally non-stationary Gaussian model is introduced via source splitting to represent the statistics of the cover image and (b)the detection of the hidden information is performed not from all but from those channels that allow to perform it with the required accuracy. We analyze the theoretically attainable bounds in such a framework and compare them to the corresponding limits of the existing state-of-the-art frameworks. The performed analysis demonstrates the superiority of the proposed approach.

Mathematical models and spatial characteristics of coherent and incoherent imaging systems

In this paper an effort of generalized approach description of image observation process was made. Block diagrams of both types of imaging systems (coherent and incoherent) are given. Peculiarities of passive imaging system permit one to formulate expressions for receiving system spatial characteristics (impulse response (blurring function) and spatial spectrum). Coherent systems were described by observation equations. Computer simulation of incoherent system spatial characteristics was performed. The necessity of image restoration for sparse antenna array was shown.

Adaptive aperture formation matched with radiometry image spatial spectrum

A new approach to radiometry image formation based on the matching of image characteristics with aperture synthesis is proposed. A quantitative analysis and comparison with the conventional method of radiometry imaging are performed.

Physical object identification using micro-structure images

A micro-structure is the common name for all fine-surface details and material properties visible when a physical object is examined at close range or under magnification. In its most basic form, the micro-structure image serves as a unique, non-cloneable identifier for that object (see Figure 1). It is non-cloneable as the current level of material science technology cannot practically produce a physical object with the precision required to clone a specific micro-structure. This protection scheme is attractive and highly competitive for large-scale, mass-market applications because of the noninvasive character of the protection and its easy, fast verification by non-experts using a mobile device. Applications include security documents, luxury items, spare aviation parts, and electronics. The non-cloneable character and uniqueness also mean that the deployed processing chain as well as the identification and authentication technologies share many elements with existing biometrics systems. We focused our micro-structure architecture elements on extracting the correct image patch, selecting robust or invariant features, dimensionality reduction, and quantization, resulting in a binary representation of the original image (see Figure 2). The acquired samples contain a printed mark used as a guide to extract the correct image patch from a fixed, determined position containing the micro-structure. This extraction needs to be vastly more precise than, for example, computer vision stitching applications. Extracted micro-patches without any geometrical distortions can be successfully modeled as Gaussian i.i.d realizations with additive white noise. This makes analytical analysis of the rest of the processing chain possible, including dimensionality reduction and quantization.1 Identification systems based on micro-structure fingerprints are elegant and fast. Figure 1. (a) A handheld mobile acquisition of a SPIE Certificate and (b-c) two extracted patches of an identical sample without any special equipment or lighting. Histogram equalization was used for visualization purposes.

Objective and subjective estimation of image restoration quality in radiometry imaging systems

In this paper two problems of radiometry imaging are considered. The first of them is related with image restoration methods used for blurred image recovery and the second one reviews the image quality criteria development.

PharmaPack: Mobile fine-grained recognition of pharma packages

We consider the problem of fine-grained physical object recognition and introduce a dataset PharmaPack containing 1000 unique pharma packages enrolled in a controlled environment using consumer mobile phones as well as several recognition sets representing various scenarios. For performance evaluation, we extract two types of recently proposed local feature descriptors and aggregate them using popular tools. All enrolled raw and pre-processed images, extracted and aggregated descriptors are made public to promote reproducible research. To evaluate the baseline performance, we compare the methods based on aggregation of local descriptors with methods based on geometrical matching.

Forensic authentication of banknotes on mobile phones.

Relatively cheap and high quality consumer scanners and printers have enabled the rise of the casual money counterfeiter. One who passes along home-made fake bills of low denomination in busy environments where the receiver is not likely to authenticate a bill. While this may be negligible on macro-economic scale, it does hurt consumers. In this paper we investigate several methods to identify counterfeit bills using an ordinary hand held mobile phone without any modification or special lighting. We demonstrate using a database of Swiss and Euro notes that variations of statistics along edges between a printing press, a laser and an inkjet are distinguishable with a mobile device. Furthermore, we show how random printing variations in the production of true banknotes can be used as a unique non-cloneable identifier for that particular bill.

Multi-class classifiers based on binary classifiers: Performance, efficiency, and minimum coding matrix distances

Using multiple binary classifiers is a popular way to construct multi-class classifiers. There exist several strategies to construct multi-class classifiers from binary classifiers. An important question is which strategy offers the highest probability of successful classification given the number of N binary classifiers used. The first result presented in this work is a method to approximate how many classes can be distinguished using N binary classifiers in practical systems rather than theoretical setups. We come to the conclusion that in this formulation, all methods share the same performance limit, which is determined using the first result. The next question is what the smallest number of binary classifiers is that is needed to attain a given probability of success. To investigate this, we introduce the concept of efficiency, which is the ratio between the number bits needed to count the number of distinguishable classes and the number of bits used. The last contribution concerns the conclusion that methods should exist that are more efficient than those currently employed.

Impulse noise removal using linear prediction model

Noise removal is an important problem in many applications. In this paper a new two-step scheme of the decision-based impulse noise removal method by means of contaminated pixel detection is proposed and comparison with direct order statistic filtering is given. The proposed methods satisfy both objective and subjective image quality.

High resolution image restoration based on inverse problem solution using iterative methods

In this paper the implementation of iterative methods to image restoration is presented. To describe the image formation process an observation model is given. Comparative analysis of different iterative image restoration methods is performed. Quantitative confirmation of image quality improvement is carried out in l/sub 2/-norm error.