Multiview stereo (MVS) aerial image depth estimation is a research frontier in the remote sensing field. Recent deep learning-based advances in close-range object reconstruction have suggested the great potential of this approach. Meanwhile, the deformation problem and the scale variation issue are also worthy of attention. These characteristics of aerial images limit the applicability of the current methods for aerial image depth estimation. Moreover, there are few available benchmark datasets for aerial image depth estimation. In this regard, this article describes a new benchmark dataset called the LuoJia-MVS dataset (https://irsip.whu.edu.cn/resources/resources_en_v2.php), as well as a new deep neural network known as the hierarchical deformable cascade MVS network (HDC-MVSNet). The LuoJia-MVS dataset contains 7972 five-view images with a spatial resolution of 10 cm, pixel-wise depths, and precise camera parameters, and was generated from an accurate digital surface model (DSM) built from thousands of stereo aerial images. In the HDC-MVSNet network, a new full-scale feature pyramid extraction module, a hierarchical set of 3-D convolutional blocks, and “true 3-D” deformable 3-D convolutional layers are specifically designed by considering the aforementioned characteristics of aerial images. Overall and ablation experiments on the WHU and LuoJia-MVS datasets validated the superiority of HDC-MVSNet over the current state-of-the-art MVS depth estimation methods and confirmed that the newly built dataset can provide an effective benchmark.

Many-objective optimization problems pose challenges to the Pareto-based multi-objective optimization algorithms. Recent studies have suggested that decomposition is a promising method to improve the performance of multi-objective evolutionary algorithms on many-objective optimization problem. Various methods based on decomposition have been developed to solve many-objective problems in recent years. However, the existing experimental comparative studies are usually limited to only a few methods based on decomposition. This paper offers a systematic comparison of seven representative decomposition-based approaches tested on two groups of widely used problems. The experimental results have demonstrated that none of the compared algorithms has a clear advantage over the others, although different algorithms are competitive on different test problems. Therefore, a careful selection of algorithms is necessary in handling a many-objective problem in hand.

Structural biology is a branch of molecular biology and biochemistry, aiming to understand the interaction of molecules like proteins by observing their structures. Crystallization is one of the most widely used methods to identify protein structure, yet it is laborious and time-consuming. Researchers are seeking assistance from computers. This paper implements an improved WDSP program for recognizing and predicting secondary structure of WD40 repeat proteins, which is a large protein family in eukaryotes. The original WDSP works well on predicting WD40 protein structures but it also suffers from low computational efficiency. We propose a more computationally efficient WDSP namely FWDSP by imposing clustering and a specific local searching to the original WDSP. Experiment results on three datasets of WD40 proteins demonstrate the effectiveness and efficiency of FWDSP.

Metabolomics data has the nature of small sample number, high dimensional, and noisy, which poses great challenges on its analysis. In this paper we propose a novel filter feature selection algorithm, namely MMAFS, for the metabolomics biomarker discovery. The MMAFS utilizes a metaheuristics chain based multimodal memetic algorithm to effectively select both local and global optimal feature subgroups that potentially contain biological meanings. A nearest-neighbor graphic based multivariate mutual information estimation is used to calculate fitness values under the max-dependency criterion. Finally, we introduce a semi-wrapper classification to improve the prediction accuracy. The MMAFS is applied on three real-world metabolomics spectrum data sets. Experimental results on 10 runs of 10-fold external cross validation show that the proposed algorithm outperforms other representative feature selection methods. Particularly, some biomarkers found by MMAFS have been proved by previous researches.

The exponential growth of high-throughput DNA sequence data has posed great challenges to genomic data storage, retrieval and transmission. Compression is a critical tool to address these challenges, where many methods have been developed to reduce the storage size of the genomes and sequencing data (reads, quality scores and metadata). However, genomic data are being generated faster than they could be meaningfully analyzed, leaving a large scope for developing novel compression algorithms that could directly facilitate data analysis beyond data transfer and storage. In this article, we categorize and provide a comprehensive review of the existing compression methods specialized for genomic data and present experimental results on compression ratio, memory usage, time for compression and decompression. We further present the remaining challenges and potential directions for future research.

Over the last few decades, metabolomics has been widely used to reveal the linkages between metabolite signal levels and physiological states. Metabolomic data are naturally high dimensional and noisy, which poses computational challenges for data analysis. In this study, a novel feature extraction method based on trimmed complex network representation is proposed for metabolomic data classification. Particularly, the proposed method begins with feature selection on the original data, and then a complex network of the selected features is constructed to represent each data sample. Afterward, the network edges are trimmed and a few topological network metrics are extracted as new features for the classification of the samples. The experimental results on a real-world metabolomic data of clinical liver transplantation demonstrate the efficiency of the proposed feature extraction method.

This paper proposes a growing partitional clustering method based on particle swarm optimization (PSO) namely PSOGC for handling data with non-spherical or non-linearly separable distribution. Particularly, PSOGC uses PSO to optimize the cluster centers. In each iteration of PSO, the particles encoding candidate cluster centers are evolved according to their social and personal knowledge. Given the candidate cluster centers, a growing strategy increasingly absorbs nearby data samples into the corresponding cluster based on k-nearest neighbor graph. The fitness of each particle is evaluated in terms of intra-cluster connectivity and inter-cluster disconnectivity of the resultant clustering. The combination of PSO and growing strategy ensures the stability of global search and the robustness of partition on data of different non-spherical shapes. Experimental results on six synthetic and three UCI real-world data sets demonstrate the efficiency of PSOGC.

SUMMARY: Experimental MS(n) mass spectral libraries currently do not adequately cover chemical space. This limits the robust annotation of metabolites in metabolomics studies of complex biological samples. In silico fragmentation libraries would improve the identification of compounds from experimental multistage fragmentation data when experimental reference data are unavailable. Here, we present a freely available software package to automatically control Mass Frontier software to construct in silico mass spectral libraries and to perform spectral matching. Based on two case studies, we have demonstrated that high-throughput automation of Mass Frontier allows researchers to generate in silico mass spectral libraries in an automated and high-throughput fashion with little or no human intervention required. AVAILABILITY AND IMPLEMENTATION: Documentation, examples, results and source code are available at http://www.biosciences-labs.bham.ac.uk/viant/hammer/.

BACKGROUND: Identifying protein-protein interactions (PPIs) is essential for elucidating protein functions and understanding the molecular mechanisms inside the cell. However, the experimental methods for detecting PPIs are both time-consuming and expensive. Therefore, computational prediction of protein interactions are becoming increasingly popular, which can provide an inexpensive way of predicting the most likely set of interactions at the entire proteome scale, and can be used to complement experimental approaches. Although much progress has already been achieved in this direction, the problem is still far from being solved and new approaches are still required to overcome the limitations of the current prediction models. RESULTS: In this work, a sequence-based approach is developed by combining a novel Multi-scale Continuous and Discontinuous (MCD) feature representation and Support Vector Machine (SVM). The MCD representation gives adequate consideration to the interactions between sequentially distant but spatially close amino acid residues, thus it can sufficiently capture multiple overlapping continuous and discontinuous binding patterns within a protein sequence. An effective feature selection method mRMR was employed to construct an optimized and more discriminative feature set by excluding redundant features. Finally, a prediction model is trained and tested based on SVM algorithm to predict the interaction probability of protein pairs. CONCLUSIONS: When performed on the yeast PPIs data set, the proposed approach achieved 91.36% prediction accuracy with 91.94% precision at the sensitivity of 90.67%. Extensive experiments are conducted to compare our method with the existing sequence-based method. Experimental results show that the performance of our predictor is better than several other state-of-the-art predictors, whose average prediction accuracy is 84.91%, sensitivity is 83.24%, and precision is 86.12%. Achieved results show that the proposed approach is very promising for predicting PPI, so it can be a useful supplementary tool for future proteomics studies. The source code and the datasets are freely available at http://csse.szu.edu.cn/staff/youzh/MCDPPI.zip for academic use.

BACKGROUND: The exponential growth of next-generation sequencing (NGS) derived DNA data poses great challenges to data storage and transmission. Although many compression algorithms have been proposed for DNA reads in NGS data, few methods are designed specifically to handle the quality scores. RESULTS: In this paper we present a memetic algorithm (MA) based NGS quality score data compressor, namely MMQSC. The algorithm extracts raw quality score sequences from FASTQ formatted files, and designs compression codebook using MA based multimodal optimization. The input data is then compressed in a substitutional manner. Experimental results on five representative NGS data sets show that MMQSC obtains higher compression ratio than the other state-of-the-art methods. Particularly, MMQSC is a lossless reference-free compression algorithm, yet obtains an average compression ratio of 22.82% on the experimental data sets. CONCLUSIONS: The proposed MMQSC compresses NGS quality score data effectively. It can be utilized to improve the overall compression ratio on FASTQ formatted files.

Omics refers to a field of study in biology such as genomics, proteomics, and metabolomics. Investigating fundamental biological problems based on omics data would increase our understanding of bio-systems as a whole. However, omics data is characterized with high-dimensionality and unbalance between features and samples, which poses big challenges for classical statistical analysis and machine learning methods. This paper studies a minimal-redundancy-maximal-relevance (MRMR) feature selection for omics data classification using three different relevance evaluation measures including mutual information (MI), correlation coefficient (CC), and maximal information coefficient (MIC). A linear forward search method is used to search the optimal feature subset. The experimental results on five real-world omics datasets indicate that MRMR feature selection with CC is more robust to obtain better (or competitive) classification accuracy than the other two measures.

This paper presents an enhanced group search optimizer (GSO), group search optimizer with intraspecific competition and lévy walk (GSOICLW), to solve the optimal power flow (OPF) problem. GSOICLW s a more biologically realistic algorithm and performs better balance between global and local searching than GSO n hat intraspecific competition IC) and lévy walk (LW) are introduced o GSO. GSOICLW is tested or the OPF problem on the IEEE 30-bus power system, with green house gases emission constraint considered. Simulation results demonstrate the accuracy and reliability of the proposed algorithm, compared with other evolutionary algorithms EAs).

Obtaining a texture-smoothing and edge-preserving filtered output is significant to image decomposition. Although the edge and the texture have salient difference in human vision, automatically distinguishing them is a difficult task, for they have similar intensity difference or gradient response. The state-of-the-art edge-preserving smoothing (EPS) based decomposition approaches are hard to obtain a satisfactory result. We propose a novel edge-preserving texture suppression filter, exploiting the joint bilateral filter as a bridge to achieve the purpose of both properties of texture-smoothing and edge-preserving. We develop the iterative asymmetric sampling and the local linear model to produce the degenerative image to suppress the texture, and apply the edge correction operator to achieve edge-preserving. An efficient accelerating implementation is introduced to improve the performance of filtering response. The experiments demonstrate that our filter produces satisfactory outputs with both properties of texture-smoothing and edge-preserving, while compared with the results of other popular EPS approaches in signal, visual and time analysis. Finally, we extend our filter to a variety of image processing applications.

This paper proposes a memory binary particle swarm optimization algorithm (MBPSO) based on a new updating strategy. Unlike the traditional binary PSO, which updates the binary bits of a particle ignoring their previous status, MBPSO memorizes the bit status and updates them according to a new defined velocity. As such, precious historical information could be retained to guide the search. The velocity vector of MBPSO is designed as a probability for deciding whether the particle bits change or not. The proposed algorithm is tested on four discrete benchmark functions. The experimental results reported over 100 runs show that MBPSO is capable of obtaining encouraging performance in discrete optimization problems.

Particle swarm optimization (PSO) is a metaheuristic algorithm that is easy to implement and performs well on various optimization problems. However, PSO is sensitive to initialization due to its rapid convergence which leads to the lack of population diversity and premature convergence. To solve this problem, a jumping-out strategy named crown jewel defense (CJD) is introduced in this paper. CJD is used to relocate the global best position and reinitializes all particles' personal best position when the swarm is trapped in local optima. Taking the advantage of CJD strategy, the swarm can jump out of the local optimal region without being dragged back and the performance of PSO becomes more robust to the initialization. Experimental results on benchmark functions show that the CJD-based PSO are comparable to or better than the other representative state-of-the-art PSO.

Gene expression data combined with clinical data has emerged as an important source for survival analysis. However, gene expression data is characterized with thousands of features/genes but only tens or hundreds of observations. The high-dimensionality and unbalance between features and samples pose big challenges for the classical survival analysis methods. This paper proposes a particle swarm optimization based radial basis function networks (PSO-RBFN) for the survival analysis on gene expression data. Particularly, PSO-RBFN applies a principle component analysis for dimensionality reduction and optimizes the RBF network using PSO. The experimental results on three gene expression datasets indicate that PSO-RBFN is able to improve the predict accuracy compared to the other classical survival analysis methods.

This paper proposes an efficient memetic clustering algorithm (MCA) for clustering based on particle swarm optimizer (PSO) and K-means. Particularly, PSO is used as a global search to allow fast exploration of the candidate cluster centers. PSO has strong ability to find high quality solutions within tractable time, but it suffers from slow-down convergence as the swarm approaching optima. K-means, achieving fast convergence to optimum solutions, is utilized as local search to fine-tune the solutions of PSO in the framework of memetic algorithm. The performance of MCA is evaluated on four synthetic datasets and three high-dimensional gene expression datasets. Comparison study to K-means, PSO, and PSO-KM (jointed PSO and K-means) indicates that MCA is capable of identifying cluster centers more precisely and robustly than the other counterpart algorithms by taking advantage of both PSO and K-means.

No-reference image quality assessment (NR IQA) has wide applicability to many problems. This paper focuses on the mechanism of divisive normalization transform (DNT) which simulates the behavior of visual cortex neurons to extract the independent components of natural images, analyzes the difference between the statistics of neighboring DNT coefficients of the images of a variety of distortion, and proposes a novel solution for NR IQA metric design. We demonstrate that measuring the statistical independence between neighboring DNT coefficients could provide features useful for quality assessment. The performance of the proposed method is quite satisfactory when it was tested on the popular LIVE, CSIQ and TID2008 databases. The experimental results are fairly competitive with the existing NR IQA metrics.

BACKGROUND: Protein-protein interactions (PPIs) play crucial roles in virtually every aspect of cellular function within an organism. Over the last decade, the development of novel high-throughput techniques has resulted in enormous amounts of data and provided valuable resources for studying protein interactions. However, these high-throughput protein interaction data are often associated with high false positive and false negative rates. It is therefore highly desirable to develop scalable methods to identify these errors from the computational perspective. RESULTS: We have developed a robust computational technique for assessing the reliability of interactions and predicting new interactions by combining manifold embedding with multiple information integration. Validation of the proposed method was performed with extensive experiments on densely-connected and sparse PPI networks of yeast respectively. Results demonstrate that the interactions ranked top by our method have high functional homogeneity and localization coherence. CONCLUSIONS: Our proposed method achieves better performances than the existing methods no matter assessing or predicting protein interactions. Furthermore, our method is general enough to work over a variety of PPI networks irrespectively of densely-connected or sparse PPI network. Therefore, the proposed algorithm is a much more promising method to detect both false positive and false negative interactions in PPI networks.

Although face recognition technology has progressed substantially, its performance is still not satisfactory due to the challenges of great variations in illumination, expression and occlusion. This paper aims to improve the accuracy of personal identification, when only few samples are registered as templates, by integrating multiple modal biometrics, i.e. face and palmprint. We developed in this paper a feature code, namely FPCode, to represent the features of both face and palmprint. Though feature code has been used for palmprint recognition in literature, it is first applied in this paper for face recognition and multi-modal biometrics. As the same feature is used, fusion is much easier. Experimental results show that both feature level and decision level fusion strategies achieve much better performance than single modal biometrics. The proposed approach uses fixed length 1/0 bits coding scheme that is very efficient in matching, and at the same time achieves higher accuracy than other fusion methods available in literature.

With the rapid development of high-throughput DNA sequencing technologies, the amount of DNA sequence data is accumulating exponentially. The huge influx of data creates new challenges for storage and transmission. This paper proposes a novel adaptive particle swarm optimization-based memetic algorithm (POMA) for DNA sequence compression. POMA is a synergy of comprehensive learning particle swarm optimization (CLPSO) and an adaptive intelligent single particle optimizer (AdpISPO)-based local search. It takes advantage of both CLPSO and AdpISPO to optimize the design of approximate repeat vector (ARV) codebook for DNA sequence compression. ARV is first introduced in this paper to represent the repeated fragments across multiple sequences in direct, mirror, pairing, and inverted patterns. In POMA, candidate ARV codebooks are encoded as particles and the optimal solution, which covers the most approximate repeated fragments with the fewest base variations, is identified through the exploration and exploitation of POMA. In each iteration of POMA, the leader particles in the swarm are selected based on weighted fitness values and each leader particle is fine-tuned with an AdpISPO-based local search, so that the convergence of the search in local region is accelerated. A detailed comparison study between POMA and the counterpart algorithms is performed on 29 (23 basic and 6 composite) benchmark functions and 11 real DNA sequences. POMA is observed to obtain better or competitive performance with a limited number of function evaluations. POMA also attains lower bits-per-base than other state-of-the-art DNA-specific algorithms on DNA sequence data. The experimental results suggest that the cooperation of CLPSO and AdpISPO in the framework of memetic algorithm is capable of searching the ARV codebook space efficiently.

This paper presents a novel affinity propagation (AP) based memetic band selection method (APMA) for hyperspectral imagery classification. The method incorporates AP based local search and genetic algorithm (GA) based global search to take advantage of both. Particularly, the AP based local search fine-tunes the GA individuals by adding relevant bands and eliminating irrelevant/redundant bands. A comparison study to the filters methods (including ReliefF, AP based method, and FCBF) and the counterpart wrapper GA feature selection on two hyperspectral imagery datasets demonstrates that APMA is capable of attaining competitive or better classification accuracy with fewer selected bands, which suggests APMA searches the band subset space more efficiently and identify better band subsets.

Due to the enormous amounts of data contained in hyperspectral imagery, the main challenge for hyperspectral image classification is to improve the accuracy with less computation complexity. Hence, dimensionality reduction (DR) is often adopted, which includes two different kinds of methods, feature extraction and feature selection. In this paper, discrete wavelet transform (DWT) and affinity propagation (AP), which belong to feature extraction and feature selection respectively, are combined together to accomplish the DR task. Firstly, DWT-based features are extracted from the original hyperspectral data; secondly, AP is applied to select representative features from the obtained ones. Experimental results demonstrate that, compared with some other DR methods which only make use of feature extraction or feature selection, the features acquired by the hybrid technique make the classification results more accurate.

A novel Fast Bacterial Swarming Algorithm (FBSA) for high-dimensional function optimization is presented in this paper. The proposed algorithm combines the foraging mechanism of E-coli bacterium introduced in Bacterial Foraging Algorithm (BFA) with the swarming pattern of birds in block introduced in Particle Swarm Optimization (PSO). It incorporates the merits of the two bio-inspired algorithms to improve the convergence for high-dimensional function optimization. A new parameter called attraction factor is introduced to adjust the bacterial trajectory according to the location of the best bacterium (bacterium with best fitness value). An adaptive step length is adopted to improve the local search ability. The algorithm has been evaluated on standard high-dimensional benchmark functions in comparison with BFA and PSO respectively. The simulation results have demonstrated the fast convergence ability and the improved optimization accuracy of FBSA.

Microarray technology enables the study of measuring gene expression levels for thousands of genes simultaneously. Cluster analysis of gene expression profiles has been applied for analyzing the function of gene because co-expressed genes are likely to share the same biological function. K-MEANS is one of well-known clustering methods. However, it requires a precise estimation of number of clusters and it has to assign all the genes into clusters. Other main problems are sensitive to the selection of an initial clustering and easily becoming trapped in a local minimum. We present a new clustering method for microarray gene data, called ppoCluster. It has two steps: 1) Estimate the number of clusters 2) Take sub-clusters resulting from the first step as input, and bridge a variation of traditional Particle Swarm Optimization (PSO) algorithm into K-MEANS for particles perform a parallel search for an optimal clustering. Our results indicate that ppoCluster is generally more accurate than K-MEANS and FKM. It also has better robustness for it is less sensitive to the initial randomly selected cluster centroids. And it outperforms comparable methods with fast convergence rate and low computation load.

A new algorithm of optimal requantization codebook design using particle-pair optimizer (PPO) is proposed to provide an effective way for image transmission over multi-speed communication system with minimal transmission delay. PPO is used for optimal codebook design in the first and second quantization respectively. In the second quantization, global distortion is used as the fitness value instead of second quantization distortion. Simulation results demonstrated that the proposed algorithm is able to achieve higher PSNR value with less transmission delay in comparison with conventional codebook optimization strategies.

This paper presents a novel genetic particle-pair optimizer (GPPO) for vector quantization of image coding. GPPO only applies a particle-pair that consists of two particles, which contributes to the relief of huge computation load in most existing vector quantization algorithms. GPPO combines the advantage both in genetic algorithms and particle swarm optimization, due to the use of genetic operators and particle operators at each generation. Experimental results have demonstrated that the quality of the codebook design optimized by GPPO is better than that optimized respectively by fuzzy K-means (FKM), fuzzy reinforcement learning vector quantization (FRLVQ, improved FRLVQ which uses fuzzy vector quantization (FVQ as post-process, called FRLVQ-FVQ, and particle-pair optimizer (PPO). GPPO provides a satisfactory solution to vector quantization, and shows a steady trend of improvement in the quality of codebook design. The dependence of the final codebook on the selection of the initial codebook is also reduced.

This paper presents a new approach toward the design of optimised codebooks by vector quantisation (VQ). A strategy of fuzzy k-means reinforced learning (FRL) is proposed which exploits the advantages offered by fuzzy clustering algorithms, competitive learning and knowledge of training vector and codevector configurations. Reinforced learning, which is consisted of attractive factor and repulsive factor, is used as a pre-process before using the conventional VQ algorithm, i.e. fuzzy k-means (FKM) algorithm. At each iteration of RL, codevectors move intelligently and intentionally toward an improved optimum codebook design. This is distinct from the standard FKM in which a random variation is introduced in the movement of the codevectors to escape from local minima. Experiments demonstrate that this results in a more effective representation of the training vectors by the codevectors and that the final codebook is nearer to the optimal solution in applications such as image compression. It has been found that the standard FKM yields improved quality of codebook design in this application when RL is used as a pre-process. The investigations have also indicated that new fuzzy k-means reinforced learning vector quantisation (FRLVQ strategy is insensitive to the selection of both the initial codebook and a learning rate control parameter, which is the only additional parameter introduced by FRL from the standard FKM.

Currently, the truncated binary exponential backoff based scheme had been defined as the mandatory contention resolution scheme in WiMAX standards. However, it had been discussed in many research that these schemes can not work perfectly with WiMAX in most cases. In this paper, a novel N modules based self-adaptive contention resolution scheme has been proposed, which divides all Subscriber Stations into several groups according to the number of available transmission opportunities in each frame and requires each Subscriber Station send bandwidth request during its group time. Meanwhile, the number of available transmission opportunities will be self-adaptively updated by Base Station according to the average collision probability presented in last frame. The analysis and simulation results show our scheme can improve the performance of whole network remarkably.

Classification of gene expression samples is a core task in microarray data analysis. How to reduce thousands of genes and to select a suitable classifier are two key issues for gene expression data classification. This paper introduces a framework on combining both feature extraction and classifier simultaneously. Considering the non-negativity, high dimensionality and small sample size, we apply a discriminative mixture model which is designed for non-negative gene express data classification via non-negative matrix factorization (NMF) for dimension reduction. In order to enhance the sparseness of training data for fast learning of the mixture model, a generalized NMF is also adopted. Experimental results on several real gene expression datasets show that the classification accuracy, stability and decision quality can be significantly improved by using the generalized method, and the proposed method can give better performance than some previous reported results on the same datasets.

A novel intelligent particle optimizer based on subvectors (IPO) is proposed in this paper, which is inspired by conventional particle swarm optimization (PSO). IPO uses only one particle instead of a particle swarm. The position vector of this particle is partitioned into a certain number of subvectors, and the updating process is based on subvectors and evolved to subvectors updating process, in which the particle adjusts the velocity intelligently by introducing a new learning factor. This learning factor utilizes the information contained in the previous updating process. The particle is capable of increasing its velocity towards the global optimum in lower dimensional subspaces and not being trapped in local optima. Experimental results have demonstrated that IPO has impressive ability to find global optimum. IPO performs better than recently developed PSO-based algorithms in solving some complicated multimodal functions.

Image watermarking has become a popular technique for authentication and copyright protection. However, many proposed image watermarking techniques are sensitive to affine transforms, such as rotation, scaling and translation. Here, a local affine transform invariant watermarking is designed and tested against attacks performed by Stirmark using the Krawtchouk moment invariants. Watermark is inserted into the perceptually significant Krawtchouk moment invariants of the original image, and watermarking based on Krawtchouk moment invariants are local, that is, the embedded watermark affects only a selected portion of the original image, the position of which can be decided by the user. This, in effect, permits the watermark to be embedded at the portion of the image which is most significant information-wise. This also means that the watermark is especially robust to cropping. Independent component analysis (ICA) is utilised by detector to extract the perfect watermark blindly. The computational aspects of the proposed watermarking are also discussed in detail. Experimental results have demonstrated that the proposed watermarking technique has a good robustness against other attacks performed by Stirmark including affine transform, cropping, filtering, image compression and random geometric distortions. It is indicated that the proposed watermarking has superior advantages over the existing ones in many aspects.