The automated detection of sea mines remains an increasingly important humanitarian and military task. In recent years, research efforts have been concentrated on developing algorithms that detect mines in complicated littoral environments. Acquired high-resolution side-looking sonar images are often heavily infested with artifacts from natural and man-made clutter. As a consequence, automated detection algorithms, designed for high probability of detection, suffer from a large number of false alarms. To remedy this situation, sophisticated feature extraction and pattern classification techniques are commonly used after detection. In this paper, we propose a nonlinear detection algorithm, based on mathematical morphology, for the robust detection of sea mines. The proposed algorithm is fast and performs well under a variety of sonar modalities and operating conditions. Our approach is based on enhancing potential mine signatures by extracting highlight peaks of appropriate shape and size and by boosting the amplitude of the peaks associated with a potential shadow prior to detection. Signal amplitudes over highlight peaks are extracted using a flat morphological top-hat by reconstruction operator. The contribution of a potential shadow to the detection image is incorporated by increasing the associated highlight amplitude by an amount proportional to the relative contrast between highlight and shadow signatures. The detection image is then thresholded at mid-gray level. The largest p targets from the resulting binary image are then labelled as potential targets. The number of false alarms in the detection image is subsequently reduced to an acceptable level by a feature extraction and classification module. The detection algorithm is tested on two side-scan sonar databases provided by the Coastal Systems Station, Panama City, Florida: SONAR-0 and SONAR-3.
KEYWORDS: Detection and tracking algorithms, Image filtering, Signal to noise ratio, Land mines, Mining, Binary data, Target detection, Multispectral imaging, Linear filtering, Reconstruction algorithms
An unsupervised algorithm is proposed for land mine detection in heavily cluttered multispectral images, based on iterating hybrid multi-spectral morphological filters. The hybrid filter used in each iteration consists of a decorrelating linear transform coupled with a nonlinear morphological detection component. Targets, extracted from the first pass, are used to improve detection results of the subsequent iteration, by helping to update covariance estimates of relevant filter variables. The procedure is stopped after a predetermined number of iterations is reached. Current implementation addresses several weaknesses associated with previous versions of the hybrid morphological approach to land mine detection. Improvement in detection accuracy and speed, robustness with respect to clutter inhomogeneity, and a completely unsupervised operation are the main highlights of the proposed approach. Our experimental investigation reveals substantially superior detection performance and lower false alarm rates over previous schemes. Properties of a graphical user interface (GUI), based on the proposed iterative morphological detection scheme, are also discussed.
KEYWORDS: Detection and tracking algorithms, Target detection, Image filtering, Reconstruction algorithms, Signal to noise ratio, Land mines, Mining, Multispectral imaging, Linear filtering, Signal detection
A new hybrid algorithm, based on combining the decorrelating and packing qualitites of Principal Component (PC) analysis and the shape extracting and filtering properties of Mathematical Morphology, is investigated in the frame-work of land mien detection. The new method is similar in spirit to the MM-MNF algorithm, which is based on a linear pre- filter, followed by a morphological multispectral detection component (MM). The new filter (PC-MM), has a similar concatenated structure, and addresses some of the weaknesses inherent in the linear component of the MM-MNF algorithm; namely, the susceptibility of the MNF transform to clutter inhomogeneity, as well as to variation sin clutter covariance estimation. The PC-MM algorithm addresses the stationarity problem by solely operating on image peaks extracted by a morphological top-hat transform. Therefore, the algorithm is much less susceptible to the present of different textural regions. Subsequently, the peaks in the extracted multispectral top-het image are projected into uncorrelated bands using the principal component (PC) transform. Due to the packing property of the PC transform, the target markers are typically found in the first and second bands in the PC transformed image. The targets are then detected using a variant of the morphological detection scheme. The new method provides a fast and satisfactory first-pass detection result, for images of different clutter homogeneities and target types. The extracted targets, from the first pass, are then issued to improve the detection result in a subsequent iteration, by updating covariance estimates of relevant filter variables.
Automatic mine detection is an area of intense research due to the implications in humanistic and battlefield management related issues. In this paper, we describe a fully automatic and iterative implementation of the nonlinear MM-MNF algorithm and review its performance for detecting landmines in multi-spectral images provided by the Coastal Battlefield Reconnaissance and Analysis program. The MM-MNF algorithm utilizes a powerful linear multi-spectral enhancement tool, called the Maximum Noise Fraction (MNF) transform, in conjunction with a nonlinear detection device based on mathematical morphology. The iterative implementation of this algorithm improves the accuracy of the clutter covariance estimation, which is turn decreases the number of false alarms, as compared to a previously reported implementation. The result are significantly better than the ones obtained from a constant false alarm rate algorithm, known as the RX-algorithm, whose performance was also inferior to the previous implementation of the MM-MNF algorithm.
The recent development of cDNA microarray allows ready access to large amount gene expression patterns for many genetic materials. Gene expression of tissue samples can be quantitatively analyzed by hybridizing fluor-tagged mRNA to targets on a cDNA microarray. Ratios of average expression level arising from co-hybridized normal and pathological samples are extracted via image segmentation, thus the gene expression pattern are obtained. The gene expression in a given biological process may provide a fingerprint of the sample development, or response to certain treatment. We propose a K-mean based algorithm in which gene expression levels fluctuate in parallel will be clustered together. The resulting cluster suggests some functional relationships between genes, and some known genes belongs to a unique functional classes shall provide indication for unknown genes in the same clusters.
A computationally efficient algorithm for computing openings by 1D flat structuring elements is proposed. The algorithm utilizes the run-length encoded image and allows implementation of the opening of a gray-scale image by a sequence of arbitrarily sized flat structuring elements. The new algorithm compares favorably to existing methods for recursive implementation of a sequence of openings, and its computation time decreases with the size of the structuring element.
Texture is an important attribute which is widely used in various image analysis applications. Among texture features, morphological texture features are least utilized in medical image analysis. From a computational standpoint, extracting morphological texture features from an image is a challenging task. The computational problem is made even greater in medical imaging applications where large images such as mammograms are to be analyzed. This paper discusses an efficient method to compute morphological texture features for any geometry of a structuring element corresponding to a texture type. A benchmarking of the code on three machines (Sun SPARC 20, Pentium II based Dell 400 workstation, and SGI Power Challenge 10000XL) as well as a parallel processing implementation was performed to obtain an optimum processing configuration. A sample processed mammogram is shown to illustrate the code outcome.
Gene expression can be quantitatively analyzed by hybridizing fluor-tagged mRNA to targets on a cDNA micro- array. Comparison of expression levels arising from co- hybridized samples is achieved by taking ratios of average expression levels for individual genes. The present paper concerns image processing to automatically segment digitized micro-arrays and measure median gene expression levees across cDNA target sites. The main difficulty arises from determination of the target site when signal intensity is low. Segmentation must be accomplished for target sites that can possess highly unstable geometry and consist of a relatively small number of pixels. Segmentation must also be computationally efficient. The present paper proposes a nonparametric statistical method that separates target site from local background using the Mann-Whitney test.
As introduced by Matheron, granulometries depend on a single sizing parameter for each structuring element. The concept of granulometry has recently been extended in such a way that each structuring element has its own sizing parameter resulting in a filter (Psi) t depending on the vector parameter t equals (t1..., tn). The present paper generalizes the concept of a parameterized reconstructive (tau) -opening to the multivariate setting, where the reconstructive filter (Lambda) t fully passes any connected component not fully eliminated by (Psi) t. The problem of minimizing the MAE between the filtered and ideal image processes becomes one of vector optimization in an n- dimensional search space. Unlike the univariate case, the MAE achieved by the optimum filter (Lambda) t is global in the sense that it is independent of the relative sizes of structuring elements in the filter basis. As a consequence, multivariate granulometries provide a natural environment to study optimality of the choice of structuring elements. If the shapes of the structuring elements are themselves parameterized, the expected error is a deterministic function of the shape and size parameters and its minimization yields the optimal MAE filter.
Assuming a random shape to be governed by a random generator and noise parameter vector, it is essential to optimally estimate the state of the generators given some set of extracted features based on the random shape. If the features used are analytically tied to shape and distortion parameters, the conditional densities involved in this Bayesian estimation problem are of a generalized nature and exist only on the manifold dictated by the particular probe. These generalized densities can be used in a conventional way to calculate the conditional- expectation estimates of the parameters. They may also be used to minimize the mean-square error on the manifold itself, thereby yielding an estimate of shape parameters consistent with the geometrical prior information provided by the observed feature set.
Assuming a random shape to be governed by a random parameter vector, a basic problem is to estimate the value of the parameter vector given some set of random features based on the random shape. The present paper considers this Bayesian estimation problem as one involving conditional densities of the random parameters conditioned by granulometric moments generated by linear granulometries. The conditional densities are interpreted as generalized functions and from these the optimal conditional-expectation estimates of the parameters given the granulometric moments are found.
As introduced by Matheron, granulometries depend on a single sizing parameter for each structuring element forming the filter. The size distributions resulting from these granulometries have been used successfully to classify texture by using as features the moments of the normalized size distribution. The present paper extends the concept of granulometry in such a way that each structuring element has its own sizing parameter and the resulting size distribution is multivariate. Classification is accomplished by taking either the Walsh or wavelet transform of the multivariate size distribution, obtaining a reduced feature set by applying the Karhunen-Loeve transform to decorrelate the Walsh or wavelet features, and classifying the textures via a Gaussian maximum-likelihood classifier.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.