Ee-Chien Chang

A fast direct Fourier-based algorithm for subpixel registration of images

This paper presents a new direct Fourier-based algorithm for performing image-to-image registration to subpixel accuracy, where the image differences are restricted to translations and uniform changes of illumination. The algorithm detects the Fourier components that have become unreliable estimators of shift due to aliasing, and removes them from the shift-estimate computation. In the presence of aliasing, the average precision of the registration is a few hundredths of a pixel. Experimental data presented here show that the new algorithm yields superior registration precision in the presence of aliasing when compared to several earlier methods and has comparable precision to the iterative method of P. Thevenaz et al. (1998).

Phase-unwrapping of SAR interferogram with multi-frequency or multi-baseline

The use of multiple InSAR images of the same area for phase unwrapping is discussed. It is shown that by combining the information in the two InSAR images, the ambiguity interval of the phase angle can be considerably lengthened, which will facilitate unwrapping. Three methods for combining multiple InSAR images are proposed and their error characteristics discussed.<<ETX>>

A wavelet approach to foveating images

Motivated by applications of foveated images in visualization, we introduce the foveation transform of an image. We study the basic properties of these transforms using the multiresolution framework of Mallat. We also consider practical methods of realizing such transforms. In particular, we introduce a new method for foveating images based on wavelets. Preliminary experimental results are shown.

Robust extraction of secret bits from minutiae

Our goal is to extract consistent bits from the same fingerprint in a noisy environment. Such bits can then be used as a secret key in several cryptographic applications. In order to correct inevitable noise during scanning and processing, a known approach extracts and publishes an additional information, known as secure sketch from the minutiae. During subsequent scanning, the sketch aids in correcting the noise to give the consistent bits. However, for minutiae (represented as 2D point set), known constructions produce sketches that are large, and are difficult to adapt to slight variations of the 2D point representation. Furthermore, even with simplified model on the noise and distribution of the minutiae, it is not clear what is the entropy of the bits extracted. To overcome the problems, we suggest using a locality preserving hash in sketch construction. We give a method that produces a small sketch and thus suitable for applications involving mobile devices. Since the sketch size is small, with a reasonable assumption, we can estimate the entropy of the secret bits extracted. In addition, we can incorporate statistical properties of the noise, and distribution of the minutiae in fine-tuning the method. Our method also includes registration of fingerprints. Experiments conducted on 4000 fingerprint images from the NIST 4 database show promising results. Assuming that an intermediate representation is uniformly distributed, with FNMR = 0.09% we are able to extract about 8 secret bits (by a conservative estimate) or 10 bits (with certain assumption on the underlying codebook).

Efficient Self-healing Key Distribution with Revocation for Wireless Sensor Networks Using One Way Key Chains

Security of group communication for large mobile wireless sensor network hinges on efficient key distribution and key management mechanism. As the wireless medium is characterized by its lossy nature, reliable communication cannot be assumed in the key distribution schemes. Therefore, self-healing is a good property for key distribution in wireless applications. The main idea of self-healing key distribution scheme is that even if during a certain session some broadcast messages are lost due to network faults, the users are capable of recovering lost session keys on their own, without requesting additional transmission from the group manager. The only requirement for a user to recover the lost session keys, is its membership in the group both before and after the sessions in which the broadcast packets containing the keys are sent. Self-healing approach of key distribution is stateless in the sense that a user who has been off-line for some period is able to recover the lost session keys immediately after coming back on-line. In this paper, we propose two constructions for scalable self-healing key distribution with trevocation capability. The novelty of our constructions are that we apply a different and more efficient self-healing mechanism compared to the ones in the literature using one-way key chain. The main improvements that our proposed schemes achieve over previous approaches are (a) communication bandwidth reduces from O((tj+ ji¾? ti¾? 1)logq) to O((t+ 1)logq), and (b) computation costs for our first and second constructions reduce from O(2tj+ j) to O(2t+ 1) and O(2(t2+ t)) respectively, where mis the maximum number of sessions, jis the current session number, tis the maximum number of compromised group members that may collude and qis a large prime number. We achieve this result without any increase in the storage complexity. The schemes are scalable to very large groups in highly mobile, volatile and hostile network. We prove in an appropriate security framework that our constructions are computationally secure and achieve both forward secrecy and backward secrecy.

Fuzzy extractors for asymmetric biometric representations

Fuzzy extractors are recently proposed error-tolerant cryptographic primitives that are potentially useful to protect biometric templates. However, there are challenges in adopting these primitives. Firstly, fuzzy extractors require the data obtained during both enrollment and verification to be in the same feature representation. However, for better performance on ROC, multiple high quality samples can be obtained during enrollment, which result in an asymmetric setting whereby data obtained in enrollment and verification are stored in different representations. Secondly, fuzzy extractors only concern about the strength of the secret key extracted, and does not directly assure that privacy is preserved. In this paper, we consider a simplified asymmetric setting and propose a sketch scheme. We analyze the key strength measured by the number of secret bits that can be extracted, and the privacy measured by the information leakage on the user identities. We next apply and investigate the scheme on FVC fingerprint datasets.

Robust, short and sensitive authentication tags using secure sketch

In order to apply cryptographic operations on noisy data, a recent approach employs some additional public data, known as secure sketch, to correct the noise so that consistent outcome can be obtained. This approach can be employed to extract authentication tags from noisy multimedia or biometric objects, by including the sketch in the tags. However, there are a few issues that need to be addressed. Firstly, those objects are typically represented in a continuous domain, and hence further quantization is required in order to obtain a short authentication tag. Secondly, for the purpose of authentication, forgery and preimage attacks are major concerns. However, such attacks are not considered in the notion of secure sketch. To handle the first issue, we give a construction using two levels of quantization. The second issue leads to the proposed additional requirement on sensitivity. We study how to choose the optimal parameters under the trade-off of robustness, size and sensitivity, and show that in many practical settings, the two-level quantization can be significantly more effective than a seemingly natural method of assigning one bit to each coefficient.

Robust image authentication using content based compression

Abstract.Image authentication is becoming very important for certifying data integrity. A key issue in image authentication is the design of a compact signature that contains sufficient information to detect illegal tampering yet is robust under allowable manipulations. In this paper, we recognize that most permissible operations on images are global distortions like low-pass filtering and JPEG compression, whereas illegal data manipulations tend to be localized distortions. To exploit this observation, we propose an image authentication scheme where the signature is the result of an extremely low-bit-rate content-based compression. The content-based compression is guided by a space-variant weighting function whose values are higher in the more important and sensitive region. This spatially dependent weighting function determines a weighted norm that is particularly sensitive to the localized distortions induced by illegal tampering. It also gives a better compactness compared to the usual compression schemes that treat every spatial region as being equally important. In our implementation, the weighting function is a multifovea weighted function that resembles the biological foveated vision system. The foveae are salient points determined in the scale-space representation of the image. The desirable properties of multifovea weighted function in the wavelet domains fit nicely into our scheme. We have implemented our technique and tested its robustness and sensitivity for several manipulations.

Watermarking color histograms

In this paper we give a method for watermarking color histograms. Color histograms have been known M. J. Swain et al., (1991) to be robust to rotations and other geometric transformations. If the watermark can be embedded in such geometry invariant representations it should survive geometric transformations. The difficulty in watermarking color histograms is that they have a nonlinear relationship with the pixel representation. Therefore it is not clear how to get a watermarked image given its watermarked histogram. We give a method for watermarking color histograms that uses earth mover distance (EMD) to modify an image to a target histogram. We conduct extensive experiments to test our method.

Subpixel registration of images

Phase correlation has been studied as a tool for registering pairs of images to subpixel resolution. Shekarforoush et al. (1994) investigated an algorithm that works well for when observed data are samples of ideally bandlimited images. In fact, images of real scenes captured by modern optics are not ideally bandlimited images. This paper examines how aliasing in sampled images effects the ideal phase relationships between continuous images undergoing translational shifts. We show that the performance of standard phase correlation methods for subpixel image registration can be severely degraded with moderate amounts of aliasing, and we investigate two approaches for modeling aliasing effects in order to improve subpixel image registration accuracy. The first approach, a very simple algorithm, both conceptually and computationally is based on detecting those frequency components that have become unreliable estimators of shift due to aliasing, and removing the detected components from the shift estimate. The second approach, more ambitious and more complex, attempts to undo the effects of aliasing and to use all de-aliased frequency components in the shift estimator.