Color Directional Local Quinary Patterns for Content Based Indexing and Retrieval
© Vipparthi and Nagar; licensee Springer 2014
Received: 26 November 2013
Accepted: 19 March 2014
Published: 2 May 2014
This paper presents a novel evaluationary approach to extract color-texture features for image retrieval application namely Color Directional Local Quinary Pattern (CDLQP). The proposed descriptor extracts the individual R, G and B channel wise directional edge information between reference pixel and its surrounding neighborhoods by computing its grey-level difference based on quinary value (−2, −1, 0, 1, 2) instead of binary and ternary value in 0°, 45°, 90°, and 135° directions of an image which are not present in literature (LBP, LTP, CS-LBP, LTrPs, DExPs, etc.). To evaluate the retrieval performance of the proposed descriptor, two experiments have been conducted on Core-5000 and MIT-Color databases respectively. The retrieval performances of the proposed descriptor show a significant improvement as compared with standard local binary pattern LBP, center-symmetric local binary pattern (CS-LBP), Directional binary pattern (DBC) and other existing transform domain techniques in IR system.
KeywordsContent based image retrieval (CBIR) Multimedia retrieval Local patterns local ternary patterns (LTP) Directional Binary Patterns (DBC)
With the radical expansion of the digitization in the living world, it has become imperative to find a method to browse and search images efficiently from immense database. In general, three types of approaches for image retrieval are, text-based, content-based and semantic based. In recent times, web-based search engines such as, Google, Yahoo, etc., are being used extensively to search for images based on text keyword searching. Here, any image needs to be indexed properly before retrieving by text-based approach. Such an approach is highly tiresome and also unrealistic to handle by human annotation. Hence, more efficient search mechanism called “content based image retrieval” (CBIR) is required. Image retrieval has become a thrust area in the field of medicine, amusement and science etc.. The search in content based approach is made by analyzing the actual content of the image rather using metadata such as, keywords, tags or descriptions associated with an image. Hence, system can filter images based on their content would provide better indexing and return more accurate results. The effectiveness of a CBIR approach is greatly depends on feature extraction, which is its prominent step. The CBIR employs visual content of an image such as color, texture, shape and faces etc., to index the image database. Hence these features can be further classified as general (texture, color and shape) and domain specific (fingerprints, human faces) features. In this paper, we mainly focused on low-level features; the feature extraction method used in this paper is an effective way of integrating low-level features into whole. Widespread literature survey on CBIR is accessible in [1–4].
The concept of color is one of the significant feature in the field of content-based image retrieval (CBIR), if it is maintained semantically intact and perceptually oriented way. In addition, color structure in visual scenery changes in size, resolution and orientation. Color histogram  based image retrieval is simple to implement and has been well used and studied in CBIR system. However, the retrieval performance of these descriptors is generally limited due to inadequacy in discrimination power mainly on immense data. Therefore, several color descriptors have been proposed to exploit special information, including compact color central moments and color coherence vector etc. reported in the literature [6, 7].
Texture is one of the most important characteristic of an image. Texture analysis has been extensively used in CBIR systems due to its potential value. Texture analysis and retrieval has gained wide attention in the field of medical, industrial, document analysis and many more. Various algorithms have been proposed for texture analysis, such as, automated binary texture feature , Wavelet and Gabor Wavelet Correlogram [9, 10], Rotated Wavelet and Rotated Complex Wavelet filters [11–13], Multiscale Ridgelet Transform  etc.. In practice texture features can be combined with color features to improve the retrieval accuracy. One of the most commonly used method is to combining texture features with color features; these include wavelets and color vocabulary trees  and Retrieval of translated, rotated and scaled color textures  etc..
In addition to the texture features, the local image features extraction attracting increasing attention in recent years. A visual content descriptor can either be local or global. A local descriptor uses the visual features of regions or objects to describe the image, where as the global descriptor uses the visual features of the whole image. Several local descriptors have been described in the literature [17–29], where the local binary pattern (LBP)  is the most popular local feature descriptor.
The main contributions of the proposed descriptor are given as follows. (a) A new color-texture descriptor is proposed, it extracts texture (DLQP) features from an individual R, G and B color channels. (b) To reduce the feature vector length of the proposed descriptor, the color-texture features were extracted from horizontal and vertical directions only.
The organization of this paper is as follows, In Section “Introduction”, introduction is presented. The local patterns with proposed descriptor are presented in Section “Local patterns with proposed Descriptor”. Section “Experimental results and discussions”, presents the retrieval performances of proposed descriptor and other state-of-the art techniques on two bench mark datasets (Corel-5000 and MIT-Color). Based on the above work Section “Conclusions” concludes this paper.
Local patterns with proposed Descriptor
Local binary patterns (LBP)
The concept of LBP was derived from the general definition of texture in a local neighborhood. This method was successful in terms of speed and discriminative performance .
Local ternary patterns (LTP)
Directional binary code (DBC)
The directional binary code (DBC) was proposed by Baochang et al. . DBC encodes the directional edge information as follows.
Color Directional Local Quinary Pattern (CDLQP)
In this section, the procedure to generate a new color-texture feature (CDLQP) descriptor is explained. Let I i be the ith plain (color space) of the image (e.g., Red color component from the “RGB” color space), where i = 1,2,3. The DLQP feature is computed independently from each (R, G and B) color channels.
For a given image I, the first-order derivatives of 0°, 45°, 90° and 135° directions are calculated using Eq. (6).
The size of the input image is N1 × N2.
In this brief, to reduce the feature vector length color-texture features were extracted from horizontal and vertical directions only.
The details of the proposed color-texture descriptor is given as follows. The steps for extracting 0° degree information is shown in Figure 1. Figure 2 and Eq. (10) explain the procedure to calculate the quinary pattern. The generated quinary pattern is further coded into two upper (A & B) and two lower (C& D) binary patterns which are shown in Figure 2. The two upper (A & B) patterns were obtained by retaining 2 by 1 and replacing 0 for −2, −1, 1 and 0 for A pattern. Likewise, pattern B was obtained by retaining 1 by 1 and replacing 0 for other values. A similar procedure was followed for other two lower patterns.
From the Figure 2, “-11, 3, −14, 8, 5, 5, −2, −4, −1” texture information are obtained when first-order derivative applied in 0° direction. Further, the derivatives are coded in to quinary pattern “-2, 1, −2, 2, 2, 2, −1, −2, −1” using upper and lower thresholds (τ1 = 2 & τ2 = 1). Finally, the quinary pattern was converted into four binary patterns (two UP and two LP). The entire operation was applied on individual color channels to generate color-texture features.
Proposed system framework for image retrieval
Figure 3 illustrates the proposed image retrieval system frame work and algorithm for the same is given below.
Algorithm: The proposed algorithm involves following steps
Separate RGB color components from an image.
Calculate the directional edge information on each color space.
Compute the local quinary value for each pixel.
Construct the CDLQP histogram for each pattern.
Construct the feature vector.
Compare the query image with images in the database using Eq. (16).
Retrieve the images based on the best matches.
Advantages of proposed methods
A new color-texture descriptor is proposed, it extracts texture (DLQP) features from an individual R, G and B color channels.
To reduce the feature vector length of the proposed descriptor, the color-texture features were extracted from horizontal and vertical directions only.
To verify the retrieval performances of CDLQP, two extensive experiments have been conducted on Corel-5000 and MIT-Color databases respectively.
The retrieval performances show a significant improvement nearly 10.78% in terms of ARP on Corel-5000 database and 9.12% improvement on MIT-Color database in terms of ARR as compared with LBP.
Experimental results and discussions
In image retrieval, various datasets are used for several purposes; these includes Corel dataset, MIT dataset and Brodtz texture dataset etc.. The Corel dataset is the most popular and commonly used dataset to test the retrieval performance, MIT-Color dataset used for texture and color feature analysis and Brodtz dataset used for texture analysis. In this paper, to verify the retrieval performances of the proposed descriptor Corel-5000 and MIT-Color datasets are used respectively.
In these experiments, each image in the database is used as the query image. The retrieval performance of the proposed method is measured in terms of recall, precision, average retrieval rate (ARR) and average retrieval precision (ARP) as given in Eq. (17) - Eq. (21) 
Where N1 is the number of relevant images (Number of images in a group), N C is a number of groups and N 2 is Total number of images to retrieve. The results obtained are discussed in the following subsections.
Experiment on Corel-5000 database
The retrieval performances of the proposed method (PM) and other existing methods on Corel-5000 database in terms of ARP and ARR
The retrieval results of the proposed method on Corel-5000 database with different distance measures in terms of ARP and ARR
Experiment on MIT-Color database
A novel evaluationary color-texture descriptor namely Color Directional Local Quinary Pattern (CDLQP) is proposed for image retrieval application. CDLQP extracts the texture features from individual R, G and B color channels using directional edge information in a neighborhood with gray-level differences between the pixels by a quinary value instead of a binary and ternary one. The extensive and comparative experiment has been conducted to evaluate our color-texture features for IR on two public natural databases namely, Corel-5000 and MIT-Color dataset. Experimental results of the proposed descriptor CDLQP show a significant improvement as compared to other state-of-the art techniques in IR system.
Santosh Kumar Vipparthi was born in 1985 in India. He received the B.E and M.Tech degrees in Electrical, Systems Engineering from Andhra University, IIT-BHU, India in 2007 and 2010 respectively. Currently he is pursuing the Ph.D. degree in the Department of Electrical Engineering at Indian Institute of Technology BHU, Varanasi, India. His major interests are image retrieval and object tracking.
Shyam Krishna Nagar was born in 1955 in India. He received the Ph.D degree in Electrical engineering from Indian Institute of Technology Roorkee, Roorkee, India, in 1991. He is currently working as Professor in Department of Electrical Engineering, Indian Institute of Technology BHU, Varanasi, Uttar Pradesh, India. His fields of interest are includes digital image processing, digital control, model order reduction and discrete event systems.
- Rui Y, Huang TS: Image retrieval: Current techniques, promising directions and open issues. J. Vis. Commun. Image Represent 1999, 10: 39–62. 10.1006/jvci.1999.0413View ArticleGoogle Scholar
- Smeulders AWM, Worring M, Santini S: A. Gupta, and R. Jain, Content-based image retrieval at the end of the early years. IEEE Trans Pattern Anal Mach Intell 2000, 22(12):1349–1380. 10.1109/34.895972View ArticleGoogle Scholar
- Kokare M, Chatterji BN, Biswas PK: A survey on current content based image retrieval methods. IETE J. Res. 2002, 48(3):261–271.View ArticleGoogle Scholar
- Ying L, Dengsheng Z, Guojun L, Wei-Ying M: A survey of content-based image retrieval with high-level semantics. Elsevier J. Pattern Recognition 2007, 40: 262–282. 10.1016/j.patcog.2006.04.045MATHView ArticleGoogle Scholar
- Swain MJ, Ballar DH: Indexing via color histograms. Proc. 3rd Int. Conf. Computer Vision, Rochester Univ., NY 1991, 11–32.Google Scholar
- Pass G, Zabih R, Miller J: Comparing images using color coherence vectors. In Proc. 4th. ACM Multimedia Conf, Boston, Massachusetts, US; 1997:65–73.Google Scholar
- Stricker M, Oreng M: Similarity of color images. In Proc. Storage and Retrieval for Image and Video Databases, SPIE; 1995:381–392.Google Scholar
- Smith JR, Chang SF: Automated binary texture feature sets for image retrieval. In Proc. IEEE Int. Conf. Acoustics, Speech and Signal Processing, Columbia Univ., New York; 1996:2239–2242.Google Scholar
- Moghaddam HA, Khajoie TT, Rouhi AH, Saadatmand MT: Wavelet Correlogram: A new approach for image indexing and retrieval. Elsevier J. Pattern Recognition 2005, 38: 2506–2518. 10.1016/j.patcog.2005.05.010View ArticleGoogle Scholar
- Moghaddam HA, Saadatmand MT: Gabor wavelet Correlogram Algorithm for Image Indexing and Retrieval. 18th Int. Conf. Pattern Recognition, K.N. Toosi Univ. of Technol., Tehran, Iran 2006, 925–928.Google Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: A Correlogram Algorithm for Image Indexing and Retrieval Using Wavelet and Rotated Wavelet Filters. International Journal of Signal and Imaging Systems Engineering 2011, 4(1):27–34. 10.1504/IJSISE.2011.039182View ArticleGoogle Scholar
- Kokare M, Biswas PK, Chatterji BN: Texture image retrieval using rotated Wavelet Filters. Elsevier J. Pattern recognition letters 2007, 28: 1240–1249. 10.1016/j.patrec.2007.02.006View ArticleGoogle Scholar
- Kokare M, Biswas PK, Chatterji BN: Texture Image Retrieval Using New Rotated Complex Wavelet Filters. IEEE Trans. Systems, Man, and Cybernetics 2005, 33(6):1168–1178. 10.1109/TSMCB.2005.850176View ArticleGoogle Scholar
- Gonde AB, Maheshwari RP, Balasubramanian R: Multiscale Ridgelet Transform for Content Based Image Retrieval. IEEE Int. Advance Computing Conf, Patial, India; 2010.View ArticleGoogle Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: Expert system design using wavelet and color vocabulary trees for image retrieval. International Journal of Expert systems with applications. 2012, 39: 5104–5114. 10.1016/j.eswa.2011.11.029View ArticleGoogle Scholar
- Cheng-Hao Y, Shu-Yuan C: Retrieval of translated, rotated and scaled color textures. Pattern Recognition 2003, 36: 913–929. 10.1016/S0031-3203(02)00124-3View ArticleGoogle Scholar
- Ojala T, Pietikainen M, Harwood D: A comparative study of texture measures with classification based on feature distributions. Pattern Recognition 1996, 29(1):51–59. 10.1016/0031-3203(95)00067-4View ArticleGoogle Scholar
- Ojala T, Pietikainen M, Maenpaa T: Multiresolution gray-scale and rotation invariant texture classification with local binary patterns. IEEE Transactions on Pattern Analysis and Machine Intelligence 2002, 24(7):971–987. 10.1109/TPAMI.2002.1017623View ArticleGoogle Scholar
- Guo Z, Zhang L, Zhang D: Rotation invariant texture classification using LBP variance with global matching. Pattern Recognition 2010, 43(3):706–719. 10.1016/j.patcog.2009.08.017MATHView ArticleGoogle Scholar
- Zhenhua G, Zhang L, Zhang D: A completed modeling of local binary pattern operator for texture classification. IEEE Transactions on Image Processing 2010, 19(6):1657–1663. 10.1109/TIP.2010.2044957MathSciNetView ArticleGoogle Scholar
- Tan X, Triggs B: Enhanced local texture feature sets for face recognition under difficult lighting conditions. IEEE Tans. Image Proc. 2010, 19(6):1635–1650. 10.1109/TIP.2010.2042645MathSciNetView ArticleGoogle Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: Directional local extrema patterns: a new descriptor for content based image retrieval. Int. J. Multimedia Information Retrieval 2012, 1(3):191–203. 10.1007/s13735-012-0008-2View ArticleGoogle Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: Local Tetra Patterns: A New Feature Descriptor for Content Based Image Retrieval. IEEE Trans. Image Process 2012, 21(5):2874–2886. 10.1109/TIP.2012.2188809MathSciNetView ArticleGoogle Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: Local Maximum Edge Binary Patterns: A New Descriptor for Image Retrieval and Object Tracking. Signal Processing 2012, 92: 1467–1479. 10.1016/j.sigpro.2011.12.005View ArticleGoogle Scholar
- Subrahmanyam M, Maheshwari RP, Balasubramanian R: Directional Binary Wavelet Patterns for Biomedical Image Indexing and Retrieval. Journal of Medical Systems 2012, 36(5):2865–2879. 10.1007/s10916-011-9764-4View ArticleGoogle Scholar
- Vipparthi S, Nagar SK: Directional Local Ternary Patterns for Multimedia Image Indexing and Retrieval. Int. J. Signal and Imaging Systems Engineering 2013, Article in: press.Google Scholar
- Takala V, Ahonen T, Pietikainen M: Block-Based Methods for Image Retrieval Using Local Binary Patterns. LNCS 2005, 3450: 882–891.Google Scholar
- Marko H, Pietikainen M, Cordelia S: Description of interest regions with local binary patterns. Pattern Recognition 2009, 42: 425–436. 10.1016/j.patcog.2008.08.014MATHView ArticleGoogle Scholar
- Baochang Z, Zhang L, Zhang D, Linlin S: Directional binary code with application to PolyU near-infrared face database. Pattern Recognition Letters 2010, 31: 2337–2344. 10.1016/j.patrec.2010.07.006View ArticleGoogle Scholar
- Available:. [Online].Google Scholar
- Available:. [Online].Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.