Advanced Discrete Cosine Transform Using Histogram Feature Selection for Face Compression Efficiency Zahraddeen Sufyanu, Member, IAENG, Dr. Fatma S. Mohamad, and Abdulganiyu A. Yusuf 1,2,3
Faculty of Informatics and Computing, 21300 Gong Badak Campus, Universiti Sultan Zainal Abidin (UniSZA), Terengganu, Malaysia 1
[email protected], 2
[email protected],
[email protected]
Abstract: Images with reduced signal level that can be easily stored in a database for recognitions are highly demanded. A system with large capacity requires a highly compressive technique for its efficiency. In this paper, more efficient image compression algorithm was disclosed, which reduced the size of original image drastically. After the necessary preprocessing, the histogram plot of an image was obtained. This histogram was rotated by 450 to produce a reduced image. Therefore, the two-dimensional Discrete Cosine Transform (2DCT) was computed on the rotated histogram to produce a new matrix. However, the proposed framework was tested on Olivetti Research Laboratory (ORL) database. And its performance was evaluated using error metrics. The new framework reported outstanding performance compared to the existing framework in the literature. Several coding approaches can be employed on the new DCT matrix for recognition. We believed that, this discovery is new in the field of signal processing.
Index Terms: Compression level; DCT; face recognition; rotated histogram 1. INTRODUCTION
I
MAGE compression is the application of data compression on digital images, meanwhile maintaining the quality of the images. The main purpose of image compression is to reduce redundancy and irrelevancy present in an image, so that it can be stored, transferred and recovered efficiently [1]. The DCT has become one of the most successful transforms in image processing for the purpose of data compression, feature extraction and recognition [2], [3]. The transform tends to concentrate the information, making it useful for image compression applications [4], [5]. It was categorized in the computer vision by Wang [6] into four slightly different transformations named DCT-I, DCT-II, DCT-III, and DCT-IV. And it has been successfully used for face recognition in [7]-[10]. Furthermore, image compression algorithms are divided into two groups: Lossless and lossy compression [11]-[13]. Lossless image compression techniques compress image without introduction of errors in the decompressed image. It does not give high compression ratio but ensures no quality loss. Conversely, in lossy image compression techniques some amount of data are discarded. Although, the compression ratio achieved is generally higher but at the expense of image quality [14]. The former occurred up to a certain level beyond which errors are introduced. All these researches and many more in the literature that used DCT especially on faces, concentrate more on extracting low frequency coefficients of the DCT matrix. Additionally, despite the energy compaction of the DCT, the acquired images may be too large when it comes to recognition of real life with many users. Hence, highly compressive technique is required even before the feature extraction. We investigated on the use of histogram recently, and the idea of rotated histogram was used on face detection using Hough transform (HT) [19]. This research brought another benefit upon Duda and Hart [20] on the restricted use of Hough transform. It also expanded the idea of using HT in image processing. To the best of the authors‟ knowledge, DCT has never been applied on the original image been compressed using histogram. In this paper, a pioneering feature extraction through DCT and histogram is proposed into the literature. The idea is to discover if DCT can be applied on a reduced feature obtained from histogram, and to find a suitable rotation angle of the histogram to obtain the appropriate feature representing original image of face. The goals of this framework are two-fold: First, to improve lossy compression techniques and promote recognition accuracy. Second, is to enable creation of a system with large number of users on a given amount of disk or memory space. As a result of that, the problem of data storage during enrollment can be minimized. In addition, the proposed algorithm is very simple to implement. The rest of the paper is organized as follows: Section II gives the concept of the proposed approach. Section III describes the materials and methods from the preprocessing stage down to the proposed feature extraction. In section IV, „Error Metrics‟ for evaluation of the proposed methodology are described. And section V presents the results and discussion, whereas in section VI concluding remarks and future study are outlined. (
)
( ) ( )
∑
∑
(
)
[
(
)
]
(
[
)
]
(1)
where √ ( )
,
√
( )
√
and √ ( The
and and
)
(
)
represent the changes induced in the spatial domain by the rotation of the histogram at angle 450. Whereas represent the reduced feature of the original image before computing the DCT.
The „M‟ and ‘N‟ are the row and column size of the rotated histogram respectively, which also varied according to change of the angle. III. MATERIALS AND METHODS A. Preprocessing Images from ORL database [21] were used to test the proposed DCT. There are 10 different images of 40 distinct subjects. Some of the subjects were taken at different times, with slightly lighting variations, facial expressions (open/closed eyes, smiling/non-smiling) and facial details (glasses/no-glasses). All the images were taken against a dark homogeneous background, and the subjects are in up-right frontal position (with tolerance for some side movement). The files are in PGM format, with a dimension of 92 x 112 per image. However, the images were converted to gray scale. Then noise was removed using median filter. In addition, mat lab R2012b version 8.0 was used. The samples of ORL face database are shown in fig. 2.
B. Histogram as a Reduction Technique The range of pixel intensities of the preprocessed images was stretched (i.e. contrast stretching), the normalized images were converted into histogram using histogram plot function, which produce distributed histograms that are easier to extract the features. However, the histogram plot was rotated by 450 to obtain the reduce-sized images of approximately squared dimensions, and that reduces complexity during recognition. Fig. 3 shows the process of histogram conversion, which is an angle-based. 𝒙
𝟐𝝅
1000
800
y
θ
𝒙
600
400
200
M
N
0 0
100
200
y
a) Histogram
b) Angle coordinate
c) Rotated histogram
Fig. 3. Histogram conversion process
i/p = X
Preprocessing
Size reduction Gray-scale Normalization Filtering
Histogram b = H( )
bins(b)
r
Rotation (45°) r = Rot(b)
Transform (2DCT) y = T(r)
Coefficient (y)
Database Face recognition 1
Quantizer c = Q(y)
c
Zig – Zag Scan + Run – level encoding
c
Entropy encoding e = E(
1
c)
e
Compressed bit stream ( , )
) Fig. 4. Proposed block diagram of the new DCT
A. Relative Reconstruction Error: We used the Tree-Structured JPEG-based DCT transform within Compressive Sensing inversion, which employs Markov chain Monte Carlo (MCMC) inference, similar to [18] to check the reconstruction error of the original images and that of equivalent rotated histogram images. Additionally, fig. 6 shows one of the results obtained, whereas table 1 summarizes these results. Fig. 7 is the representation of the results in line graph acquired through 40 iterations using 32 x 32 images. The reconstruction errors were plotted against number of subjects, with the results arranged in ascending order to simply check the errors variations. The error in the new DCT is quite lower than the existing DCT using the original image.
V.
Results and discussion fig. 8 displays three results of the new transform. The first row are the normalized images obtained using stretchlim () function at 0.05
and 0.95 (minimum and maximum) pixel points. The second row shows histogram plots of the normalized images obtained through imhist() function. And the third row displays the equivalent rotated histograms obtained through imrotate() function at an angle of 450. Whereas the fourth row are the corresponding DCT matrices obtained through the dct2() function. TABLE II COMPARISON BETWEEN EXISTING AND NEW DCT USING ERROR METRICS Methods
Avg. MSE
Avg. PSNR (DB)
Avg. RSME
Avg. Mean Absolute Error
EXISTING DCT NEW DCT
0.991
52.978
0.995
1.430
0.154
61.069
0.392
0.211
Original Image
Original Sparse Coefficients 1500 1000
10
500
20
0 30 10
20
30
200
Reconstructed Sparse Coefficients 1500 1000
400
600
800
1000
Recovered Image, rela err = 0.072361
10
500
20
0 30 200
400
600
800
1000
10
20
30
Fig. 8. Some of the results of the new DCT applied on the rotated histogram Fig. 6. Results of 2-dimensional 32 x 32 images acquired using the original image
The rotation was tested from 00 at a step of 150 until an approximately squared image of symmetrical structure was obtained at 450. The relative reconstruction error significantly reduced in the rotated histogram image compared to that of original image. Furthermore, the following reasons describe the advantages of compressing the original image in this paper: The size of the rotated histogram is 2.6KB while that of the original image is 7.2KB. This signifies the optimized data. Moreover, each matrix shows significant decorrelation with respect to one another. Therefore, even extracting the 8 x 8 blocks as shown in Fig. 5 will produce appropriate feature matrix during training, let alone few blocks from these coefficients. This proves that, the efficiency of image compression algorithm depends on how best and swift the redundancies can be exploited, which may be through reduction of image size without any appreciable loss of quality [14]. The results in fig. 7 demonstrate lower relative reconstruction error in the new DCT compared to the existing one. In the same vein, table 2 concluded that, DCT can be applied on a reduced feature obtained from histogram, at a specified angle with lesser MSE values and higher PNSR values.
VI. Conclusion and subsequent study A new feature extraction frame work using DCT through histogram was introduced into the literature. The aim for improving the compression ability of the DCT using the reduced feature obtained from histogram was achieved. The paper advances the research in this field by applying the DCT on the extracted features of faces. It was reported that, the relative reconstruction error of the rotated histogram representing the face was amply reduced compared to the errors encountered in the original image. Hence, this concept optimized the compression level and efficiency in lossy compression techniques. In addition, it creates massive storage ability on a given amount of disk or memory space; this is the uniqueness of the study. The resulted matrix showed significant decorrelation with other images. However, the performance of the proposed algorithm was evaluated within a CS inversion algorithm of the Markovian structure used in [18] and other compression parameters.
Acknowledgment Z. Sufyan thanks his mentor Bashir Muhammad, Department of Electrical Engineering, Universiti of Technology Malaysia (UTM). REFERENCES [1] [2] [3] [4] [5] [6] [7]
R. Nitu, B. Savita, “Comparative Analysis of Image Compression Using DCT and DWT Transforms,” International Journal of Computer Science and Mobile Computing, vol. 3 no. 7, pp. 990 – 996. 2014. X. Jing, D. Zhang, “A Face and Palmprint Recognition Approach Based on Discriminant DCT Feature Extraction,” IEEE Transaction on Systems, Man and Cybernetics-Part B: Cybernetics, vol. 34 no. 6, 2004. T. Amornraksa, S. Tachaphetpiboon, “Fingeprint Recognition using DCT Feature,” Electronic Letters, vol. 42 no. 9, 2006. Jain, Anil K., Fundamentals of Digital Image Processing, Englewood Cliffs, NJ, Prentice Hall, 1989, pp. 150-153. Pennebaker, William B., and Joan L. Mitchell, JPEG: Still Image Data Compression Standard, Van Nostrand Reinhold, 1993. Z Wang, “Fast Algorithms for the Discrete Wavelet Transform and for the Discrete Fourier Transform,” IEEE Trans. Acoust., Speech, and Signal Proc., vol. 32, pp. 803–816. 1984. W. C. Kao, M. C. Hsu and Y. Y. Yang, “Local Contrast Enhancement and Adaptive Feature Extraction for Illumination-invariant Face Recognition”, Pattern Recognition, vol. 43, no. 5, pp. 1736-1747. 2010.
[8] S. Dabbaghchian, M. P. Ghaemmaghami, A. Aghagolzadeh, “Feature Extraction using Discrete Cosine Transform and Discrimination Power Analysis with a Face Recognition Technology,” Pattern Recognition, vol. 43 no. 4, pp.1431-1440. 2010. [9] P. Cui, R. B. Zhang, “A semi-supervised Coefficient Select Method for Face Recognition,” Journal of Harbin Engineering University, vol. 33, no. 7 pp. 855-861. 2012. [10] P. Cui, R. B. Zhang, “Face Feature Extraction Method Based on Part of Labeled Data”, Journal of Optoelectronics Laser, vol. 23, no. 3, pp. 554-560. 2012. [11] Pratt and K. William, Digital Image Processing. [12] R. C. Gonzalez and R. E. Woods, Digital Image Processing, Pearson Education, 2002. [13] D. Saloman, Data Compression. [14] S. Dutta, A. Abhinav, P. Dutta, P. Kumar, and A. Halder, “ An Efficient Image Compression Algorithm Based on Histogram Based Block Optimization and Arithmetic Coding, ” International Journal of Computer Theory and Engineering, vol. 4, no. 6, pp. 954 – 957, Dec. 2012. [15] S. M. Fatma, A. Azizah, & C. Suriayati, “Histogram Matching For Color Detection: A Preliminary Study,” IEEE,978-1-4244-6716-7/10/$26.00 2010. [16] Z. M. Hafed and M. D. Levine, "Face recognition using the discrete cosine transfonn," International Journal of Computer Vision, vol. 43, no.3, pp.167·188, 2001. [17] P. J. S. G. Ferreira and A. J. Pinho, “ Why Does Histogram Packing Improve Lossless Compression Rates?,” IEEE Signal Processing Letters, vol. 9, no. 8, pp. 259-261, Aug. 2002. [18] L. He, H. Chen, and L. Carin,”Tree-Structured Compressive Sensing with Variational Bayesian Analysis” IEEE Signal Processing Letters, vol. 17, no. 3, pp. 233-236, March 2010. [19] S. Zahraddeen, S. M. Fatma, A. Y. Abdulganiyu, A. S. Ben-musa, “A new Hough transform on face detection using histograms,” IVCJ Image and vision computing journal by Elsevier, submitted for publication. [20] Duda, R. O. and P. E. Hart. "Use of the Hough Transformation to Detect Lines and Curves in Pictures," Comm. ACM, Vol. 15, pp. 11–15. 1972. [21] Orldatabase: Available: http://www.cl.cam.ac.uk/research/dtg/ attarchive/ facedatabase. html [22] H. K. Ekenel and R. Stiefelhagen, "Local appearance based face recognition using discrete cosine transform, " In 13th European Signal Processing Conference, Antalya, Turkey, 2005. [23] S. Zahraddeen, S. M. Fatma, A. Y. Abdulganiyu, “an efficient discrete cosine transform and gabor filter based feature extraction for face recognition,” The 6th International Conference on Postgraduate Education (ICPE-6 2014), UTeM 17th-18thDecember 2014, to be published.
