International Journal of Computational Intelligence and Applications

A retina has a space-variant sampling mechanism and an orientation-sensitive mechanism. The space-variant sampling mechanism of the retina is called Retinotopic Sampling (RS). With these mechanisms, the object-detection is formulated as finding an appropriate coordinate transformation from a coordinate system on the input image to the retina. The appropriate coordinate transformation is found using maximum likelihood method. By using the model based on RS, we formulate a kernel function as an analytical function of the information on the input image, the position and the size of the object in the input image. Then the object-detection is realised as a gradient decent method for a discriminant function trained by Support Vector Machine (SVM). This detection mechanism realises faster detection than exploring a visual scene in raster-like fashion. The discriminant function outperforms results of SVMs using a kernel function using intensities of all pixels (based on independently published results), in face detection experiments over test images in the MIT-CBCL face database.

Segmentation of natural textures has been investigated by developing a novel semi-supervised support vector machines (S³VM) algorithm with multiple constraints. Unlike conventional segmentation algorithms the proposed method does not classify the textures but classifies the uniform-texture regions and the regions of boundaries. Also the overall algorithm does not use any training set as used by all other learning algorithms such as conventional SVMs. During the process, the images are restored from high spatial frequency noise. Then various-order statistics of the textures within a sliding two-dimensional window are measured. K-mean algorithm is used to initialise the clustering procedure by labelling part of the class members and the classifier parameters. Therefore at this stage we have both the training and the working sets. A non-linear S³VM is then developed to exploit both sets to classify all the regions. The convex algorithm maximises a defined cost function by incorporating a number of constraints. The algorithm has been applied to combinations of a number of natural textures. It is demonstrated that the algorithm is robust, with negligible misclassification error. However, for complex textures there may be a minor misplacement of the edges.

We propose a keyword specification method for images, which can be used to retrieve an image by a keyword. In order to specify a keyword for a sub-region of the image, images are segmented in some regions. Here, we consider ten keywords to specify the regions. The image segmentation method consists of the maximum-distance algorithm, labeling, and merging the small regions. We provide training regions for each keyword. Important features of the keyword are selected using the Factor Analysis (FA). The features are compressed into a two-dimensional space using a Sandglass-type Neural Network (SNN). We train the SNN using the training dataset. Then 60 samples of the testing dataset are evaluated by keywords extraction.

Research on "man-machine interface" has increased in many fields of engineering and its application to facial expressions recognition is expected. The eigenface method by using the principal component analysis (PCA) is popular in this research field. However, it is not easy to compute eigenvectors with a large matrix if the cost of calculation when applying it for time-varying processing is taken into consideration.

In this paper, in order to achieve high-speed PCA, the simple principal component analysis (SPCA) is applied to compress the dimensionality of portions that constitute a face. A value of cos θ is calculated using an eigenvector by SPCA as well as a gray-scale image vector of each picture pattern. By using neural networks (NNs), the difference in the value of cos θ between the true and the false (plastic) smiles is clarified and the true smile is discriminated. Finally, in order to show the effectiveness of the proposed face classification method for true or false smiles, computer simulations are done with real images. Furthermore, an experiment using the self-organisation map (SOM) is also conducted as a comparison.

A digital watermarking scheme based on vector quantisation (VQ) for gray watermark is proposed. It begins with the procedure of encoding the gray watermark, and embeds the encoded indices into the cover image in VQ domain then. To improve the performance of the watermarking scheme, a genetic index assignment (GIA) procedure, which modifies the signal of the watermark to suit the signal of the cover image, is proposed. The proposed gray watermark embedding scheme is easy to implement, requires no original cover image to be presented during extraction, expands the size of the used watermark, and provides better watermarked results. Experimental results will show the novelty and effectiveness of it.

We propose a robust digital watermarking technique based on Principal Component Analysis (PCA) and evaluate the effectiveness of the method against some watermark attacks. In this proposed method, watermarks are embedded in the PCA domain and the method is closely related to DCT or DWT based frequency-domain watermarking. The orthogonal basis functions, however, are determined by data and they are adaptive to the data. The presented technique has been successfully evaluated and compared with DCT and DWT based watermarking methods. Experimental results show robust performance of the PCA based method against most prominent attacks.

We propose an approach for real-time blind source separation (BSS), in which the observations are linear convolutive mixtures of statistically independent acoustic sources. A recursive least square (RLS)-like strategy is devised for real-time BSS processing. A normal equation is further introduced as an expression between the separation matrix and the correlation matrix of observations. We recursively estimate the correlation matrix and explicitly, rather than stochastically, solve the normal equation to obtain the separation matrix. As an example of application, the approach has been applied to a BSS problem where the separation criterion is based on the second-order statistics and the non-stationarity of signals in the frequency domain. In this way, we realise a novel BSS algorithm, called exponentially weighted recursive BSS algorithm. The simulation and experimental results showed an improved separation and a superior convergence rate of the proposed algorithm over that of the gradient algorithm. Moreover, this algorithm can converge to a much lower cost value than that of the gradient algorithm.

Although several highly accurate blind source separation algorithms have already been proposed in the literature, these algorithms must store and process the whole data set which may be tremendous in some situations. This makes the blind source separation infeasible and not realisable on VLSI level, due to a large memory requirement and costly computation. This paper concerns the algorithms for solving the problem of tremendous data sets and high computational complexity, so that the algorithms could be run on-line and implementable on VLSI level with acceptable accuracy. Our approach is to partition the observed signals into several parts and to extract the partitioned observations with a simple activation function performing only the "shift-and-add" micro-operation. No division, multiplication and exponential operations are needed. Moreover, obtaining an optimal initial de-mixing weight matrix for speeding up the separating time will be also presented. The proposed algorithm is tested on some benchmarks available online. The experimental results show that our solution provides comparable efficiency with other approaches, but lower space and time complexity.