Smith J. R., & Chang S.F. ( May 1996). Automated binary texture feature sets for image retrieval. Proceedings of the IEEE. ICASSP-96. Atlanta, GA.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... The system is capable of performing a wide variety of complex queries due to an efficient indexing, and also because spatial issues such as adjacency, overlap and encapsulation can be addressed by the system [56] The retrieval process is accentuated by using binary tree based indexing algorithms [44, 52, 53, 55]. Contrary to VisualSEEk, WebSEEk [57] is a catalog based search engine for the World Wide Web. WebSEEk supports queries on both subject catalog and visual properties such as color, spatial layout and texture matching features [44] The system collects images using several autonomous Web spiders, ....

J. R. Smith and S.-F. Chang. Automated binary texture feature sets for image retrieval. In n Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, pages 2239--2242, Atlanta, GA, USA, May 1996.

....by its moments. Therefore, a 9 dimensional color feature vector (3 moments for each color channel, HSV) is extracted and stored from every image in the database. Texture refers to the visual pattern with properties of homogeneity that do not result from the presence of a single color or intensity [3]. It contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment. To represent texture, we used the wavelet texture representation proposed by Smith and Chang in [4] Specifically, we feed an image into a wavelet filter bank and ....

J. R. Smith and S. F. Chang, "Automated Binary Texture Feature Sets for Image Retrieval", Proc. IEEE Intl. Conf. Acoust., Speech, and Signal Proc., Atlanta, GA, 1996.

.... measure of similarity provides an accurate and efficient measure of similarity between two images based on their color [42] Texture Features: Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [45]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. Texture contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [23] Because of its importance 4 IEEE TRANSACTIONS ON ....

 J.R. Smith and S.-F. Chang, "Automated Binary Texture Feature Sets for Images Retrieval," Proc. ICASSP, Atlanta, 1996.

....feature used in [1] and [2] to some extent, are based on models of human texture perception. More recently, several random field based texture models and multi scale filtering methods [6, 11] have been studied. Use of texture for content based retrieval has been explored by several researchers [3, 7, 8]. Among these, features computed from Gabor filtered images appear quite promising. A comprehensive evaluation on using Gabor features can be found in [7, 9] More recent evaluation and comparison using other texture features also support the observation that the orientation and scale selective ....

John R. Smith and Shih-Fu Chang, "Automated binary texture feature sets for image retrieval", Proc. ICASSP-96, Atlanta, GA, 1996.

....deviation) and third moments of the H, S and V coordinates of each pixel are computed and then normalized, a standard Euclidean distance is used. Texture Features refer to the visual patterns that have homogeneity properties that do not result from the presence of a single color or intensity [29]. Texture is an innate property of virtually all surfaces, it contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [9] Because of its importance and usefulness in Pattern Recognition and Computer Vision, extensive ....

J. R. Smith and S.-F. Chang. Automated Binary Texture Feature Sets for Image Retrieval. In Proc ICASSP-96, Atlanta, GA, 1996.

.... based measure of similarity provides an accurate and efficient measure of similarity between two images based on their color [42] Texture Features: Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [45]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. Texture contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [23] Because of its importance and usefulness in ....

John R. Smith and Shih-Fu Chang. Automated Binary Texture Feature Sets for Image Retrieval. In Proc ICASSP-96, Atlanta, GA, 1996.

....Color Sets and the conventional Color Histogram was further discussed in [Smith and Chang, 1995a, Smith and Chang, 1995b] 2. 2 Texture Features Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [Smith and Chang, 1996]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. It contains 3 important information about the structural arrangement of surfaces and their relationship to the surrounding environment [Haralick et al. 1973] Because of its importance and ....

.... was introduced and its theoretical framework established, many researchers began to study its applications to texture representation [Smith and Chang, 1994, Chang and Kuo, 1993, Laine and Fan, 1993, Gross et al. 1994, Kundu and Chen, 1992, Thyagarajan et al. 1994] In [Smith and Chang, 1994, Smith and Chang, 1996] Smith and Chang used the mean and variance statistics extracted from the Wavelet subbands as the texture representation. This approach achieved over 90 accuracy on the 112 Brodatz texture images. 2.3 Shape Features In general, the shape representations can be divided into two categories, ....

Smith, J. R. and Chang, S.-F. (1996). Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA.

....Image Content (QBIC) system being developed at IBM Almaden Research Center (Faloutsos et al. 1993; Flickner etal 1995) The QBIC system supports queries based on color, texture, sketch, and layout of images. Another important related project is the ADVENT system developed at the Columbia University (Smith and Chang, 1994, 1995, 1996; Wang 1995) Their main research focus is color texture region extraction in both the uncompressed domain and the compressed domain. The color set concept is used in their color region extraction approach to make it faster and more robust. Its texture region extraction is based on the features ....

Smith John R. and Chang, Shih-Fu (1996). Automated Binary Texture Feature Sets for Images Retrieval, In Proc. ICASSP.

....to compute distance between color moments of two images. The distance is then converted into similarity value [19] Co occurrence Matrix Texture [12] represents visual patterns in an image. It has properties of homogeneity that do not result from the presence of only a single color or intensity [28]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. Euclidean distance function is used to obtain similarity between two co occurrence feature vectors. Textual keywords are extracted in addition to image visual content from the textual ....

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA, May 1996.

....Content #QBIC# system being developed at IBM Almaden Research Center #Faloutsos et al. 1993; Flickner etal 1995#. The QBIC system supports queries based on color, texture, sketch, and layout of images. Another important related project is the ADVENT system developed at the Columbia University #Smith and Chang, 1994, 1995, 1996; Wang 1995#. Their main research focus is color#texture region extraction in both the uncompressed domain and the compressed domain. The color set concept is used in their color region extraction approach to make it faster and more robust. Its texture region extraction is based on the features ....

Smith John R. and Chang, Shih-Fu #1996#. Automated Binary Texture Feature Sets for Images Retrieval, In Proc. ICASSP.

....a 5x5 window. The third, and final, step stores the transformed image in a quad tree structure. Texture feature extraction requires three steps. The first step transforms the 64 blocks of 16x16 pixels in to 64 sets of wavelet data using a Quadratic Mirror Filter (QMF) 2 iterations, 7 subbands) [6] Each wavelet data produces seven subbands of means and variances; i.e. a 14 element vector. In the second step, the texture vectors are then compared to reference VisTex textures [7] in the known texture table to generate 64 texture indices representing textures for blocked data. The third, and ....

J. R. Smith and S.-F. Chang, Automated Binary Texture Feature Sets for Image Retrieval, International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 1996

....each pixel are computed and then normalized resulting in a vector of 9 elements. The distance function used is Euclidean distance. Texture Features: Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [46]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. Texture contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [20] Because of its importance and usefulness in ....

....engine, both of which are developed at Columbia University. The main features are spatial relationship queries of image regions and visual feature extraction from compressed domain [56, 8, 9, 10] The visual features used in their systems are Color Set and Wavelet Transform based texture feature [44, 45, 43, 46]. To speed up the retrieval process, they also developed binary tree based indexing algorithms [47, 11, 12, 49] VisualSEEk supports queries based on both visual features and their spatial relationships. This enables a user to submit a sunset query as red orange color region on top and blue or ....

John R. Smith and Shih-Fu Chang. Automated Binary Texture Feature Sets for Image Retrieval. In Proc ICASSP-96, Atlanta, GA, 1996.

....of DWT coefficients are compared for image indexing. The comparison starts at the lowest resolution, and the results are progressively refined using higher resolution DWT coefficients. This technique fails to retrieve images even with small camera operations or object motions. Smith and Chang [13] have recently proposed a technique for automatic extraction of texture regions in the wavelet domain. Here, the DWT coefficients are first thresholded based on their energy and a bi level image is generated corresponding to each subimage. A morphological operator is then applied on the bi level ....

....analysis are presented in section 3 and 4, respectively. Performance of the proposed techniques is presented in section 5, which is followed by conclusions. 2. Wavelet Histogram Technique In this section, we provide a brief review of the wavelet histogram technique proposed by Smith and Chang [13]. Here, the image is first decomposed to M stages using wavelets. The wavelet bands form a pyramid of M levels, where the level k bands are the highpass bands after the k th stage decomposition. A three stage DWT decomposition is shown in Fig. 2, where level 1 bands consists of 9 8 7 , A A A ....

[Article contains additional citation context not shown here]

J. R. Smith and S. F. Chang, "Automated binary texture feature sets for image retrieval," Proc. of ICASSP, Vol. 4, pp. 2239-2242, Atlanta, May 1996.

....subband coefficients are employed as index. However, the indexing performance of these techniques is not robust to translation and rotation that are crucial in imaging applications. Indexing techniques based on directional information, or band statistics have been proposed by Smith et al. [24]. Here, a nine channel (for three stage decomposition) thresholded texture image is created from the highpass DWT coefficients of each image. A 512 bin texture histogram is generated by considering pixels from all channels corresponding to each spatial location. This technique provides good ....

J. R. Smith and S. F. Chang, "Automated binary texture feature sets for image retrieval," Proc. of IEEE Int. Conf. on Acoust., Speech, and Signal Processing Atlanta, (May 1996).

.... based measure of similarity provides an accurate and efficient measure of similarity between two images based on their color [38] Texture Features: Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [41]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. Texture contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [20] Because of its importance and usefulness in ....

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA, 1996.

....The topics are 2D and 3D multiresolution image segmentation, e.g. for ultrasonic images. Other application include tissue characterisation which is also done in agricultural inspection by Kim et al. [18] Applications of texture features for searches in large image databases are shown by Smith [30] [29] and in [24] Combined colour and texture descriptions are expected to become very important in this area. Another successful area is remote sensing, where promising work was reported by Clausi [6] Other applications are found in material science, where characterisation of corrosion is ....

J. R. Smith and S. Chang. Automated binary texture feature sets for image retrieval. In Proc. Int. Conf. Acoust. Speech and Signal Processing, Atlanta, GA, 1996.

....using an 8 Theta 8 2D histogram over the HS coordinates and the V coordinate is dropped since it is easily affected by the lighting condition. The histogram intersection measure of distance provides an accurate and efficient measure of (dis)similarity between two images based on their color [9]. Texture Features: represented via a modified version of CCD (coarseness, contrast, and directionality) developed in [5] The Coarseness measures granularity of the texture (fine vs coarse) and is represented This work was supported in part by the Army Research Laboratory under Cooperative ....

....fuzzy model. This may be attributed to more information being preserved in the ranking functions used for combination based on and and or operations. 5. Conclusions Most existing content based image retrieval systems also extract low level image features like color, texture, shape, and structure [5, 1, 6, 9]. These systems, however, support queries on single features separately which we refer to as simple queries which limits their usefulness to end users. In contrast, similar to the approaches taken in information retrieval system, the approach we have taken in developing MARS is to support an ....

J. R. Smith and S.-F. Chang, "Automated binary texture feature sets for image retrieval," in Proc ICASSP-96.

.... regions and visual feature extraction from compressed domain [161, 26, 27, 29] The visual features used in their systems are Color Set and Wavelet Transform based texture feature [138, 139, 137, 140] To speed up the retrieval process, they also developed binary tree based indexing algorithms [135, 28, 141, 142]. VisualSEEk supports queries based on both visual features and their spatial relationships. This enables a user to submit a sunset query as red orange color region on top and blue or green region at the bottom as its sketch . WebSEEk is a web oriented search engine. It consists of three main ....

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc. IEEE Int. Conf. Acoust., Speech, and Signal Proc., May 1996.

....fast search. The relationship between the proposed Color Sets and the conventional Color Histogram was further discussed [138, 139] 2.2. Texture Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [140]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. It contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [59] Because of its importance and usefulness in Pat tern ....

....[43] and MARS system [63, 101] further improved this texture representation. In early 90 s, after Wavelet transform was introduced and its theoretical framework established, many researchers began to study the use of Wavelet transform in texture representation [137, 30, 72, 53, 71, 158] In [137, 140], Smith and Chang used the statistics (mean and variance) extracted from the Wavelet subbands as the texture representation. This approach achieved over 90 accuracy on the 112 Brodatz texture images. To explore the middle band characteristics, treestructured Wavelet transform was used by Chang ....

[Article contains additional citation context not shown here]

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc. IEEE Int. Conf. Acoust., Speech, and Signal Proc., Atlanta, GA, 1996.

.... regions and visual feature extraction from compressed domain [162, 27, 28, 29] The visual features used in their systems are Color Set and Wavelet Transform based texture feature [139, 140, 138, 141] To speed up the retrieval process, they also developed binary tree based indexing algorithms [136, 30, 142, 143]. VisualSEEk supports queries based on both visual features and their spatial relationships. This enables a user to submit a sunset query as red orange color region on top and blue or green region at the bottom as its sketch . WebSEEk is a web oriented search engine. It consists of three main ....

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc. IEEE Int. Conf. Acoust., Speech, and Signal Proc., May 1996.

....fast search. The relationship between the proposed Color Sets and the conventional Color Histogram was further discussed [139, 140] 2.2. Texture Texture refers to the visual patterns that have properties of homogeneity that do not result from the presence of only a single color or intensity [141]. It is an innate property of virtually all surfaces, including clouds, trees, bricks, hair, fabric, etc. It contains important information about the structural arrangement of surfaces and their relationship to the surrounding environment [59] Because of its importance and usefulness in Pattern ....

....system [44] and MARS system [64, 102] further improved this texture representation. In early 90 s, after Wavelet transform was introduced and its theoretical framework established, many researchers began to study the use of Wavelet transform in texture representation [138, 31, 73, 54, 72, 159] In [138, 141], Smith and Chang used the statistics (mean and variance) extracted from the Wavelet subbands as the texture representation. This approach achieved over 90 accuracy on the 112 Brodatz texture images. To explore the middle band characteristics, treestructured Wavelet transform was used by Chang ....

[Article contains additional citation context not shown here]

John R. Smith and Shih-Fu Chang. Automated binary texture feature sets for image retrieval. In Proc ICASSP-96, Atlanta, GA, 1996.

No context found.

Smith J. R., & Chang S.F. ( May 1996). Automated binary texture feature sets for image retrieval. Proceedings of the IEEE. ICASSP-96. Atlanta, GA.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC