This paper conducts a survey of modern binary pattern flavored feature extractors applied to the Facial Expression Recognition (FER) problem. In total, 26 different feature extractors are included, of which six are se...
详细信息
ISBN:
(纸本)9781479981748
This paper conducts a survey of modern binary pattern flavored feature extractors applied to the Facial Expression Recognition (FER) problem. In total, 26 different feature extractors are included, of which six are selected for in depth description. In addition, the paper unifies important FER terminology, describes open challenges, and provides recommendations to scientific evaluation of FER systems. Lastly, it studies the facial expression recognition accuracy and blur invariance of the Local Frequency Descriptor. The paper seeks to bring together disjointed studies, and the main contribution is to provide a solid overview for future research.
Measuring image quality is an interesting and challenging area of research. In this paper we investigate the performance of the statistical functions called copula as image quality measures. These functions are popula...
详细信息
ISBN:
(纸本)9781479900312
Measuring image quality is an interesting and challenging area of research. In this paper we investigate the performance of the statistical functions called copula as image quality measures. These functions are popular for applications where data distributions are unknown. This property motivated some researchers to using these copulas in imageprocessing in general and in detecting image changes and image registration in particular. In this research, we use the Gaussian copula to calculate the mutual information, which is the measure of the association of the reference and the distorted or tampered with images. To test the performance of the proposed method, we implemented our method on LIVE image database and compared our results with three popular image quality measures namely Visual Information Fidelity (VIF), Structural Similarity (SSIM), and Universal Quality Measure (UQI). The results show that our quality measure, obtained similar results to the three methods in 99% of the time, hence the proposed method can be considered as an efficient image quality index.
In this paper, we present an image watermarking scheme based on discrete wavelet transform and singular value decomposition (SVD) for color images. This scheme embeds the watermark into the luminance (Y) channel of YC...
详细信息
ISBN:
(纸本)9780889869264
In this paper, we present an image watermarking scheme based on discrete wavelet transform and singular value decomposition (SVD) for color images. This scheme embeds the watermark into the luminance (Y) channel of YC bCR color space of the host image. The luminance image is decomposed up to four DWT levels. SVD is applied on the HL1-4 and LH1-4sub-bands of each level and pseudo random number (PRN) sequence is added to the singular values of their coefficients to produce what we call modified singular values. To obtain the modified HL and LH coefficients, we take the inverse of the SVD of the modified SVs. Finally, we perform the IDWT on the modified HL & LH coefficients as well as HH and LL coefficients to produce the watermarked Y channel image. Then, the modified luminance, Cb and Cr are combined to get the watermarked color image. Simulation results show that our watermark is robust against high JPEG compression, filtering, Gaussian-noise, resizing and other imageprocessing attacks. In addition, this watermark is not susceptible to any false positive or detection of false reference watermark.
Architectural elements are the components and details of buildings. Their unique set, combination, design, construction technique form the architectural style of buildings. Building facade classification by architectu...
详细信息
Architectural elements are the components and details of buildings. Their unique set, combination, design, construction technique form the architectural style of buildings. Building facade classification by architectu...
详细信息
Architectural elements are the components and details of buildings. Their unique set, combination, design, construction technique form the architectural style of buildings. Building facade classification by architectural styles is viewed as a task of classifying separate architectural structural elements. In the scope of building facade architectural style classification the current paper targets the problem of classification of Gothic and Baroque architectural elements called tracery, pediment and balustrade. Since certain gradient directions dominate on the shape of each architectural element, discrimination between dominating gradients means classification of architectural elements and thus architectural styles. We use local features to describe gradient directions. Our approach is based on clustering and learning of local features and yields a high classification rate.
Building facade classification by architectural styles allows categorization of large databases of building images into semantic categories belonging to certain historic periods, regions and cultural influences. image...
详细信息
作者：
Alzahir, SaifImage Processing
Graphics and Multimedia Lab Computer Science Department UNBC Prince George BC V2N 4Z9 Canada
In this paper we propose a model-based binary image compression scheme. In this scheme, we merge one-dimensional (1-D) blocks of black pixels of the input binary image with those in consecutive rows into larger blocks...
详细信息
作者：
Alzahir, SaifGraphics
Image Processing and Multimedia Lab Computer Science University of Northern British Columbia Prince George BC V2N 4Z9 Canada
In recent years, an unparalleled volume of textual information was transported over the Internet via email, chatting, blogging, twittering, digital libraries, and information retrieval systems. As the volume of text d...
详细信息
作者：
Saif alZahirImage Processing
Graphics and Multimedia Lab Computer Science Department UNBC Prince George BC Canada
In this paper we propose a model-based binary image compression scheme. In this scheme, we merge one-dimensional (1-D) blocks of black pixels of the input binary image with those in consecutive rows into larger blocks...
详细信息
In this paper we propose a model-based binary image compression scheme. In this scheme, we merge one-dimensional (1-D) blocks of black pixels of the input binary image with those in consecutive rows into larger blocks using mathematical models that preserve the quality of the image. This process reduces the number of vertices when the image is segmented into rectangles for compression. The top-left and the bottom-right vertices of each generated rectangle are then identified and the coordinates of which are efficiently encoded. The model for merging the blocks was obtained through extracting the data values involving the blocks of various widths and the subjective tolerance of an average viewer. The data values are then plotted on a Cartesian plane and approximated with linear, logarithmic, and polynomial functions. Simulation results show that the images after merging using the proposed model have less number of rectangles without any obvious image distortion and have higher compression ratio those rectangular partitioning methods in the literature.
作者：
Saif alZahirGraphics
Image Processing and Multimedia Lab Computer Science University of Northern British Columbia Prince George BC Canada
In recent years, an unparalleled volume of textual information was transported over the Internet via email, chatting, blogging, twittering, digital libraries, and information retrieval systems. As the volume of text d...
详细信息
In recent years, an unparalleled volume of textual information was transported over the Internet via email, chatting, blogging, twittering, digital libraries, and information retrieval systems. As the volume of text data has exceeded 40% of the total volume of traffic on the Internet, compressing textual data becomes imperative. Many algorithms were introduced and employed for this purpose including Huffman encoding, arithmetic encoding, the Ziv-Lempel family, Dynamic Markov Compression, and Burrow-Wheeler Transform. In this paper, a novel algorithm for compressing textual images is presented. The algorithm comprises of two parts: (i) a fixed-to-variable codebook; and (ii) row and column reduction coding scheme, RCRC. Simulation results on a large number of Arabic textual images show that this algorithm has a compression ratio of approximately 87%, which exceeds published results including those of JBIG2.
暂无评论