Current Issue : January - March Volume : 2016 Issue Number : 1 Articles : 5 Articles
Because the reliability of feature for every pixel determines the accuracy of classification, it is important to design a\nspecialized feature mining algorithm for hyperspectral image classification. We propose a feature learning algorithm,\ncontextual deep learning, which is extremely effective for hyperspectral image classification. On the one hand, the\nlearning-based feature extraction algorithm can characterize information better than the pre-defined feature\nextraction algorithm. On the other hand, spatial contextual information is effective for hyperspectral image\nclassification. Contextual deep learning explicitly learns spectral and spatial features via a deep learning architecture\nand promotes the feature extractor using a supervised fine-tune strategy. Extensive experiments show that the\nproposed contextual deep learning algorithm is an excellent feature learning algorithm and can achieve good\nperformance with only a simple classifier....
The purpose of this paper is to improve the robustness of traditional image watermarking based on singular value\ndecomposition (SVD) by using optimization-based quantization on multiple singular values in the wavelet domain.\nIn this work, we divide the middle-frequency parts of discrete-time wavelet transform (DWT) into several square\nblocks and then use multiple singular value quantizations to embed a watermark bit. To minimize the difference\nbetween original and watermarked singular values, an optimized-quality formula is proposed. First, the peak\nsignal-to-noise ratio (PSNR) is defined as a performance index in a matrix form. Then, an optimized-quality\nfunctional that relates the performance index to the quantization technique is obtained. Finally, the Lagrange\nPrinciple is utilized to obtain the optimized-quality formula and then the formula is applied to watermarking.\nExperimental results show that the watermarked image can keep a high PSNR and achieve better bit-error rate\n(BER) even when the number of coefficients for embedding a watermark bit increases....
While face analysis from images is a well-studied area, little work has explored the dependence of facial appearance\non the geographic location from which the image was captured. To fill this gap, we constructed GeoFaces, a large\ndataset of geotagged face images, and used it to examine the geo-dependence of facial features and attributes, such\nas ethnicity, gender, or the presence of facial hair. Our analysis illuminates the relationship between raw facial\nappearance, facial attributes, and geographic location, both globally and in selected major urban areas. Some of our\nexperiments, and the resulting visualizations, confirm prior expectations, such as the predominance of ethnically\nAsian faces in Asia, while others highlight novel information that can be obtained with this type of analysis, such as\nthe major city with the highest percentage of people with a mustache...
Anatomical structures and tissues are often hard to be segmented in medical images due to their poorly defined\nboundaries, i.e., low contrast in relation to other nearby false boundaries. The specification of the boundary polarity\ncan help alleviate a part of this problem. In this work, we discuss how to incorporate this property in the relative fuzzy\nconnectedness (RFC) framework. We include a theoretical proof of the optimality of the new algorithm, named\noriented relative fuzzy connectedness (ORFC), in terms of an oriented energy function subject to the seed constraints,\nand show its usage to devise powerful hybrid image segmentation methods. The methods are evaluated using\nmedical images of MRI and CT of the human brain and thoracic studies....
Spiking neural networks (SNN) have gained popularity in embedded applications such as robotics and computer\nvision. The main advantages of SNN are the temporal plasticity, ease of use in neural interface circuits and reduced\ncomputation complexity. SNN have been successfully used for image classification. They provide a model for the\nmammalian visual cortex, image segmentation and pattern recognition. Different spiking neuron mathematical\nmodels exist, but their computational complexity makes them ill-suited for hardware implementation. In this paper, a\nnovel, simplified and computationally efficient model of spike response model (SRM) neuron with spike-time\ndependent plasticity (STDP) learning is presented. Frequency spike coding based on receptive fields is used for data\nrepresentation; images are encoded by the network and processed in a similar manner as the primary layers in visual\ncortex. The network output can be used as a primary feature extractor for further refined recognition or as a simple\nobject classifier. Results show that the model can successfully learn and classify black and white images with added\nnoise or partially obscured samples with up to Ã?â??20 computing speed-up at an equivalent classification ratio when\ncompared to classic SRM neuron membrane models. The proposed solution combines spike encoding, network\ntopology, neuron membrane model and STDP learning....
Loading....