Scale-invariant Image Research Articles

Robust FFT-Based Scale-Invariant Image Registration with Image Gradients

We present a robust FFT-based approach to scale-invariant image registration. Our method relies on FFT-based correlation twice: once in the log-polar Fourier domain to estimate the scaling and rotation and once in the spatial domain to recover the residual translation. Previous methods based on the same principles are not robust. To equip our scheme with robustness and accuracy, we introduce modifications which tailor the method to the nature of images. First, we derive efficient log-polar Fourier representations by replacing image functions with complex gray-level edge maps. We show that this representation both captures the structure of salient image features and circumvents problems related to the low-pass nature of images, interpolation errors, border effects, and aliasing. Second, to recover the unknown parameters, we introduce the normalized gradient correlation. We show that, using image gradients to perform correlation, the errors induced by outliers are mapped to a uniform distribution for which our normalized gradient correlation features robust performance. Exhaustive experimentation with real images showed that, unlike any other Fourier-based correlation techniques, the proposed method was able to estimate translations, arbitrary rotations, and scale factors up to 6.

Robust Feature Based Image Watermarking Process

A digital image watermarking scheme must be robust against a variety of possible attacks. In the proposed approach, a new rotation and scaling invariant image watermarking scheme is proposed based on rotation invariant feature and image normalization. The rotation invariant features are extracted from the segmented areas and are selected as reference points. Sub-regions centered at the feature points are used for watermark embedding and extraction. Image normalization is applied to the sub-regions to achieve scaling invariance. In the scheme, first, the image is segmented into a number of homogeneous regions and the feature points are extracted. Then the circular regions for watermark embedding or extraction are defined. Based on the image normalization and orientation assignment, the rotation, scaling, and translation invariant regions can be used for watermark embedding and extraction. The segmented image is modeled as mixture generalized Gaussian distribution and this model is the basis of mathematical analysis of various aspects of the watermarking processes such as probability of error, embedding strength adjustment. The watermark embedding strength is adjusted adaptively using the noise visibility function.The original image is not needed for the watermark detection. The effectiveness and accuracy of the proposed scheme is established through experimental results.

RST Invariant Image Watermarking Algorithm With Mathematical Modeling and Analysis of the Watermarking Processes

In this paper, a new rotation and scaling invariant image watermarking scheme is proposed based on rotation invariant feature and image normalization. A mathematical model is established to approximate the image based on the mixture generalized Gaussian distribution, which can facilitate the analysis of the watermarking processes. Using maximum a posteriori probability based image segmentation, the cover image is segmented into several homogeneous areas. Each region can be represented by a generalized Gaussian distribution, which is critical for the analysis of the watermarking processes mathematically. The rotation invariant features are extracted from the segmented areas and are selected as reference points. Sub-regions centered at the feature points are used for watermark embedding and extraction. Image normalization is applied to the sub-regions to achieve scaling invariance. Meanwhile, the watermark embedding and extraction schemes are analyzed mathematically based on the established mathematical model. The watermark embedding strength is adjusted adaptively using the noise visibility function and the probability of error is analyzed mathematically. The mathematical relationship between fidelity and robustness is established. The experimental results show the effectiveness and accuracy of the proposed scheme.

Rotation, scaling and translation invariant image watermarking using feature points

In this article, a novel robust image watermarking scheme is presented to resist rotation, scaling, and translation (RST). Initially, the original image is scale normalized, and the feature points are then extracted. Furthermore, the locally most stable feature points are used to generate several nonoverlapped circular regions. These regions are then rotation normalized to generate the invariant regions. Watermark embedding and extraction are implemented in the invariant regions in discrete cosine transform domain. In the decoder, the watermark can be extracted without the original image. Simulation results show that the proposed scheme is robust to traditional signal processing attacks, RST attacks, as well as some combined attacks.

Learning How to Extract Rotation-Invariant and Scale-Invariant Features from Texture Images

Learning how to extract texture features from noncontrolled environments characterized by distorted images is a still-open task. By using a new rotation-invariant and scale-invariant image descriptor based on steerable pyramid decomposition, and a novel multiclass recognition method based on optimum-path forest, a new texture recognition system is proposed. By combining the discriminating power of our image descriptor and classifier, our system uses small-size feature vectors to characterize texture images without compromising overall classification rates. State-of-the-art recognition results are further presented on the Brodatz data set. High classification rates demonstrate the superiority of the proposed system.

A survey of RST invariant image watermarking algorithms

In this article, we review the algorithms for rotation, scaling and translation (RST) invariant image watermarking. There are mainly two categories of RST invariant image watermarking algorithms. One is to rectify the RST transformed image before conducting watermark detection. Another is to embed and detect watermark in an RST invariant or semi-invariant domain. In order to help readers understand, we first introduce the fundamental theories and techniques used in the existing RST invariant image watermarking algorithms. Then, we discuss in detail the work principles, embedding process, and detection process of the typical RST invariant image watermarking algorithms. Finally, we analyze and evaluate these typical algorithms through implementation, and point out their advantages and disadvantages.

Phase Fourier vector model for scale invariant three-dimensional image detection

A scale invariant 3D object detection method based on phase Fourier transform (PhFT) is addressed. Three-dimensionality is expressed in terms of range images. The PhFT of a range image gives information about the orientations of the surfaces in the 3D object. When the object is scaled, the PhFT becomes a distribution multiplied by a constant factor which is related to the scale factor. Then 3D scale invariant detection can be solved as illumination invariant detection process. Several correlation operations based on vector space representation are applied. Results show the tolerance of detection method to scale besides discrimination against false objects.

A New Technique for Digital Image Watermarking

In this paper, a new technique is proposed for rotation, scaling and translation (RST) invariant image watermarking based on log-polar mappings (LPM) and phase-only filtering (POF). The watermark is embedded in the LPM of Fourier magnitude spectrum of the original image, and a small portion of resulting LPM spectrum is used to calculate the watermark positions. This technique avoids computing inverse log-polar mapping (ILPM) to preserve the quality of the watermarked image, and avoids exhaustive search to save computation time and reduce false detection. Experimental results demonstrate that the digital watermarking technique is invariant and robust to rotation, scaling, and translation transformation.

RST invariant digital image watermarking based on a new phase-only filtering method

Watermarked image may undergo geometric transforms which make watermark detection difficult or impossible. This paper presents a new rotation, scaling, and translation (RST) invariant digital image watermarking scheme based on log-polar mapping and a new phase-only filtering method. Watermark is embedded in the spatial domain imperceptibly based on the human perceptual model. Because of the log-polar mapping, the rotation and scaling in the spatial domain result in shifts in the log-polar mapping domain. Based on the new phase-only filtering method, the shift parameters in the log-polar mapping domain can be computed, and then the rotation degree and scaling ratio in the spatial domain can be successfully obtained through inverse log-polar mapping computation. The translation parameters in the spatial domain can also be calculated through the phase-only filtering method. After obtaining the RST parameters, the watermarked image can be transformed back to its original orientation, size, and position. Then the watermark can be detected easily and accurately. This scheme does not need an original image to detect the watermark. Only two templates derived from the original image are needed for the watermark detection. The size of each template is 64 × 64 bytes and can be compressed to 1 K bytes without loss of computation accuracy. Experiments show that the proposed scheme is robust against various attacks including geometric attacks. The method of inverting RST transforms can also be used in other watermarking algorithms.

Mobile Robot Localization and Mapping with Uncertainty using Scale-Invariant Visual Landmarks

A key component of a mobile robot system is the ability to localize itself accurately and, simultaneously, to build a map of the environment. Most of the existing algorithms are based on laser range finders, sonar sensors or artificial landmarks. In this paper, we describe a vision-based mobile robot localization and mapping algorithm, which uses scale-invariant image features as natural landmarks in unmodified environments. The invariance of these features to image translation, scaling and rotation makes them suitable landmarks for mobile robot localization and map building. With our Triclops stereo vision system, these landmarks are localized and robot ego-motion is estimated by least-squares minimization of the matched landmarks. Feature viewpoint variation and occlusion are taken into account by maintaining a view direction for each landmark. Experiments show that these visual landmarks are robustly matched, robot pose is estimated and a consistent three-dimensional map is built. As image features are not noise-free, we carry out error analysis for the landmark positions and the robot pose. We use Kalman filters to track these landmarks in a dynamic environment, resulting in a database map with landmark positional uncertainty.

Rotation, scale and translation invariant spread spectrum digital image watermarking

A digital watermark is an invisible mark embedded in a digital image which may be used for a number of different purposes including image captioning and copyright protection. This paper describes how a combination of spread spectrum encoding of the embedded message and transform-based invariants can be used for digital image watermarking. In particular, it is described how a Fourier–Mellin-based approach can be used to construct watermarks which are designed to be unaffected by any combination of rotation and scale transformations. In addition, a novel method of CDMA spread spectrum encoding is introduced which allows one to embed watermark messages of arbitrary length and which need only a secret key for decoding. The paper also describes the usefulness of Reed Solomon error-correcting codes in this scheme.

An FFT-based technique for translation, rotation, and scale-invariant image registration

This correspondence discusses an extension of the well-known phase correlation technique to cover translation, rotation, and scaling. Fourier scaling properties and Fourier rotational properties are used to find scale and rotational movement. The phase correlation technique determines the translational movement. This method shows excellent robustness against random noise.

An Artificial Compound Eye Using a Microlens Array and Its Application to Scale-Invariant Processing

An artificial compound eye corresponding to apposition eyes of insects has been demonstrated using a two-dimensional array of gradient-index rod lenses known as a Selfoc lens plate (a SLP). A single erect Moire image is reconstructed from a number of inverted micro-images by an experimental optical system composed of the SLP, a pinhole array and a plano-concave lens. Characteristics of the compound eye is briefly discussed from the viewpoints of space-bandwidth product, viewing angle, and focal depth. Capability of a scale-invariant image sensor that has been proposed as an application of the compound-eye-type imager was also proved experimentally.

Scale-invariant image recognition based on higher-order autocorrelation features

We propose a framework and a complete implementation of a translation and scale-invariant image recognition system for natural indoor scenes. The system employs higher-order autocorrelation features of scale space data which permit linear classification. An optimal linear classification method is presented, which is able to cope with a large number of classes represented by many, as well as very few samples. In the course of the analysis of our system, we examine which numerical methods for feature transformation and classification show sufficient stability to fulfill these demands. The implementation has been extensively tested. We present the results of our own application and several classification benchmarks.

Scale, rotation, and shift invariant image recognition

A circular harmonic filter illuminated with white light illumination is used to achieve scale, rotation, and shift invariant image recognition. The circular harmonic expansion of an object is utilized to achieve rotation and shift invariant image recognition. Scale invariance is added using a broadband dispersion-compensation technique. When illuminated with broadband light, the frequency-plane filter simply selects the wavelength from the spectrum that produces the same size transform as that recorded in the filter. Laboratory experiments that demonstrate the operation of the technique are presented.

Scale-invariant image processing by means of scaled transforms or form-invariant, linear shift-variant filters

Scaled transforms have recently been introduced as a powerful mathematical tool for processing (e.g., detecting) images that are arbitrarily scaled or rotated. It will be shown that the scaled transforms are equivalent to a subclass of the linear shift-variant, "form-invariant" filters. The equivalence holds in the sense that, given any scaled transform and any input signal, a linear shift-variant filter can be found such that the outputs of the filter and of the scaled transform are Fourier-transform pairs. The relevance of this result is briefly discussed.