An enhancement method on illumination images - Akademia Baru

Journal of Advanced Review on Scientific Research 28, Issue 1 (2016) 33-41

Penerbit

Akademia Baru

Journal of Advanced Review on Scientific Research Journal homepage: www.akademiabaru.com/arsr.html ISSN: 2289-7887

An enhancement method on illumination images: A survey

Open Access

Wan Azani Mustafa 1,*, Haniza Yazid 1, Noratikah Mazlan 2 1 2

School of Mechatronic Engineering, University Malaysia Perlis, 02600 Ulu Pauh, Arau, Perlis, Malaysia School of Engineering Technology, Kampus Sg. Chuchuh, Universiti Malaysia Perlis, 02100 Padang Besar, Perlis

ARTICLE INFO

ABSTRACT

Article history:

Image enhancement has found to be one of the most important vision applications because it has the ability to enhance the visibility of images. Contrast variables and uneven illumination are considered one of the most challenging tasks in the image enhancement field. The badly contrast images commonly caused by occlusion, pose illumination which is highly non-linear. Distinctive procedures have been proposed so far for improving the quality of the digital images. To enhance picture quality image enhancement can specifically improve and limit some data presented in the input picture. It is a kind of vision system which reductions picture commotion, kill antiquities, and keep up the informative parts. Its object is to open up certain picture characteristics for investigation, conclusion, and further use. The main objective of this paper is to discover the limitations of the existing image enhancement strategies. Here in this paper, a survey of all the techniques related to contrast enhancement of images is given. Also, the paper contains all the advantages and disadvantages of these techniques.

Received 11 November 2016 Received in revised form 15 December 2016 Accepted 16 December 2016 Available online 25 December 2016

Keywords: Enhancement, Background, Contrast, Luminosity, Survey

Copyright © 2016 PENERBIT AKADEMIA BARU - All rights reserved

1. Introduction The current era of computer science and medical science functioning together to solve complex challenges such as diseases and medical instruments which are highly effective to use for the solution of algorithms and mathematical problems, while some challenges still need to improve their techniques and algorithms, for instance, facial recognition, voice recognition, and processing of a language are current issues [1]. Pre-processing is an important phase in image analysis [2]. Usually, due to the non-ideal acquisition process, the acquired images tend to have non-uniform illumination. Non-uniform illumination can come from different sources such as aging filaments and contaminated apertures [3-4]. So, before the image of the retinal is transmitted to the doctor for analysis, the luminosity and contrast variation should be eliminated to show the right information about the *

Corresponding author. E-mail address: [email protected] (Wan Azani Mustafa)

33

Penerbit

Journal of Advanced Review on Scientific Research Volume 28, Issue 1 (2016) 33-41

Akademia Baru

condition of the patient [5]. In the image processing field, the illumination problem is a big factor which needs to be considered before segmentation process. Now, the luminosity and contrast normalization become a challenging task for researchers. Thus, various methods have been proposed to solve this problem. Illumination correction can be categorized into two main techniques; prospective correction and retrospective correction. A lot of researchers proposing a new method for illumination and contrast correction techniques based on retrospective correction since it produces a better result compared to the prospective correction [6-7]. Fig. 1 shows the contrast variation problem in a retinal image from Digital Retinal Images for Vessel Extraction (DRIVE) database.

Contrast variation problem

Fig. 1. The contrast variation problem in retinal images.

2. Common problem Recent trends show that the usage of image processing is becoming more and more prominent in our daily life. In addition to television, camera and personal computer, many high-tech electronic products, such as handphone, or even a refrigerator, nowadays are being equipped with capabilities to display digital images. Unfortunately, the input images that are captured by (or provided for) these devices are sometimes not really in good brightness and contrast. Therefore, a process known as a digital image contrast enhancement is normally required to increase the quality of these low contrast images. The retinal image may encounter variation in intensity which occurred due to general imperfection during the image acquisition process [8]. This inhomogeneous brightening over the retinal gives a visibility problem for the lesion in the images. This is a serious problem to automate the abnormalities in the retinal images. Correction of non-uniform illumination actually is very important in image processing applications such as quantitative analysis, registration, and segmentation [9]. The digital retinal camera produces a high-quality color digital image and instantly displays it on a large, high-resolution monitor, however, the light reflected by the retinal surface produce a luminosity and contrast variation [10]. Then, before the image of the retinal is transmitted to a doctor for analysis, this luminosity and contrast variation should be eliminated to show the right information about the condition of the patient. Therefore, the pre-processing stage is crucial to overcoming this problem. Retinal images are acquired using a fundus camera always have a luminosity and contrast problem [11]. In the retinal image, there are various types of lesions such as exudates, microaneurysms, hemorrhages, optic disc, fovea, and blood vessel which is the indicators for several diseases such as diabetic retinopathy, neovascularization, and glaucoma. The non-uniform illuminated retinal images were obscure the visibility of the lesions and can lead to incorrect

34


Penerbit

Akademia Baru

diagnosis. The corrected retinal images were transmitted to the expert/ophthalmologist either for evaluation or diagnosis. 2. Background correction methods 2.1. Retinal images Usually, due to the non-ideal acquisition process, the acquired images tend to have non-uniform illumination [12]. The luminosity and contrast estimation in the background image is an important part that must be considered in order to normalize the illumination in the retinal image. A lot of researchers try to find the best solutions to solve this problem. Gonzales and Wood normalized a luminosity image by disposing of low-frequency luminosity drifts by means of high-pass filtering, or by approximating the drift with some mathematical function and then subtracting this component of the observed image [8-9]. Besides, Øien and Osnes applied a large median filter to extract slow variations of luminosity, which were then subtracted from the observed image [15]. These techniques have been highlighted in the specific application such as for retinal images. Local adaptive non-linear filters are studied by Wallis, to produce a better local contrast and automatically improved the image quality. However, the drawbacks were decreased the global contrast, where the difference between bright and dark features. The techniques do not guarantee the reduction of luminosity variation throughout the image [16]. In order to enhance the illumination variation, B-spline approximation method was reported by Kolar et al. [17]. The control points for the B-splines approach are determined from the original image, separately to the 3 channels called as red, green, and blue. The above finding is consistent with the study by Kubecka et al. They presented a new method based on the B-spline shading model for non-uniform illumination modification following Shannon's entropy and applying two models; multiplicative and parametric local bias. Then, the estimated illumination surface is used in a multiplicative model for fast correction to find the final corrected images. In 2005, Forancchia et al. have proposed a method in order to solve the illumination problem in the retinal images based on statistical analysis [10]. The method was obtained the luminosity and contrast drift as the illumination models. The image correction was done by using a mathematical algorithm. The experiment result was shown in Fig. 2. In order to improve the illumination correction in retinal image application, Zheng et al. suggested a novel method based on the gradient distribution in frequency property and it can determine the illumination inhomogeneity automatically by considering the Gradient Distribution Sparsity and Parametric Model of Bias Field [6].

Fig. 2. The retinal image; (Left) original greyscale and (Right) result of the Forancchia method [10]

In retinal image application, Grisan et al. [18] discovered a method based on hue, saturation, and value (HSV). A new algorithm was proposed to estimate the illumination and finally, corrected the

35

Penerbit


Akademia Baru

images. The evaluation of Shannon’s entropy is done based on the Parzen windowing method with the spline-based shading model. Mora et al. suggested a novel simple method in order to normalize the uneven illumination using smoothing spline technique. The chosen smoothing spline is a special class of spline that can capture the low frequencies that characterize the non-uniform illumination [19]. The resulting performance efficient compared to conventional methods. In 2015, Azani et al. explained a new method based on a combination of low pass filter and Gaussian filtering [20]. Based on the SNR value, the proposed approach is effective to eliminate the illumination and reducing the contrast variation effects. Besides, Azani et al. presented a comprehensive review for illumination correction based on six (6) types of filtering such as Low pass filter, High pass filter, and Homomorphic filter [21]. The result of comparison techniques was illustrated in Fig. 3.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 3. The resulting image after applied six (6) type enhancement methods; (a) the original, (b) greyscale image, (c) low pass filter, (d) high pass filter, (e) high boost filter, (f) Homomorphic low pass filter, (g) Homomorphic high pass filter, and (h) Homomorphic high boost filter [21].

2.2. Other applications Based on the reviews, many researchers have argued that the homomorphic filtering effectively to improve and enhance the illumination and contrast problem [22–26]. The homomorphic filtering technique can increase the contrast of original images by adjusting the brightness across an image and simultaneously normalizes the image [24-27]. Fig. 4 shows a basic flow for homomorphic filtering. A combination of filtering types also can be used for illumination normalization had been suggested by Zhu et al. which combine the low-pass and homomorphic filter before proposed a new algorithm model to estimate the local light of the background and regularize the image contrast variation [28].

Fig. 4. The basic flow of homomorphic filter [21].

36


Penerbit

Akademia Baru

In a bad illumination image, Fan and Zhang describe a homomorphic filtering technique based on a combination of the histogram equalization (HE) and the Difference of Gaussian (DoG) filter for contrast enhancement. The resulting performance is effective because it has not only correcting the illumination, however, also preserve the edge and details in the entire images [29]. A lot of researchers used the Homomorphic filtering to solve the illumination problem, however, there are two main problems; it doesn’t provide an indication of the cut-off frequency (CF) and introduces another artifact of the inflammation in the foreground border. So, Ardizzone et al. proposed a new technique based on the quantitative evaluation to solve this problem [30]. This main purpose of this technique is to find the accurate a cut-off automatically. Tao and Veldhuis introduced a method based on horizontal Gaussian derivative filters and local binary patterns. This method is suitable and efficient especially for 2D illumination images [31]. Actually, the illumination normalization is a very important process before the segmentation process, hence Li et al. proposed a combination algorithm between ICR (Illumination Compensation based on Multiple Regression Model) and histogram equalization (HE) to solve the illumination problem [32]. A fast illumination normalization for strong shadow was studied by Park et al. [33]. This technique consists two steps: (1) regularize and approximate the illumination with 3 × 3 and (2) iteratively convolving and then dividing the original input image with the estimated value. The result images show the contrast enhanced. In 2008, Salah-Eldin et al. presented a new method to solve the illumination problem in face recognition based on Histogram Equalization (HE), Log Transformation (LOG), Gamma Correction, and The Compression Function of the Retinal Filter (COMP) [34]. This technique called as GAMMAHM-COMP involved three stages: (1) applying the gamma correction on the reference average image, (2) histogram matching the input image to the result from stage 1, and (3) applying the Retinal filter’s compression function to further enhancing the final result. Vishwakarma et al. proposed Discrete Cosine Transform (DCT) low-frequency coefficients by using histogram equalization (HE) to stretch the intensity before applying the DCT algorithm and this approach are fast and easy to apply in real time system [35]. The research study by Lian et al. also found that DCT able to solve the illumination and contrast problem. They proposed a method based on noise estimation which is obtained from high frequency using Discrete Cosine Transform (DCT) coefficients was assumed as an illumination model, finally, the illumination was eliminated and enhanced [36]. Kyung and Kyu introduced adaptive smoothing filter based on iterative convolution and two discontinuity measures; spatial gradient and local inhomogeneity for illumination correction [37]. The above findings contradict the study by Cheng et al. [38]. They were presented quotient image and different smoothing filter techniques to eliminate the bad image under varying contrast condition and this method are effective and efficient compared to Morphological Quotient Image (MQI), SelfQuotient Image (SQI) and Dynamic Morphological Quotient Image (DMQI) [38]. They also proposed a new model based on 2D Gaussian illumination for correcting the contrast variation using the Quadtree technique to locate the dark area [39]. A new algorithm based on Fuzzy C-Means (FCM) was presented by Vlachos and Dermatas to reduce the effect of illumination image. This technique based on applying the inverse of the image formation by replicating FCM convergence and, finally the contrast variation was corrected [9]. In 2012, a simple method for illumination correction was investigated by Leahy et al. by obtaining the illumination profile using multiplicative image formation models and Laplace interpolation [8]. Wang et al. proposed a method based on Weber’s law to eliminate the unwanted light and produce a good contrast variation [40]. This method involved two steps; first obtaining the intensity variation using Laplace operator and second, applying an algorithm from Lambertian reflectance to find the

37

Penerbit


Akademia Baru

final result. Lee and Kim focused on nonlinear minimum squares to measure the binaries of the image and overlap the illumination on the bi-level image [41]. Additionally, they also recommend retrospective correction techniques to regularize the contrast variation and illumination. A normal distribution (NORM), histogram equalization (HE), histogram matching (HM) and Gamma Intensity Correction (GIC) are four parameters that need to be determined before applying local normalization methods which had been studied by Santamar and Palacios [42]. They suggested using a window size in order to increase the result performance. The suggested method called as Local Histogram Equalization (LHE), Local Histogram Matching (LHM), and Local Normal Distribution (LNORM). In this experiment, the error rate was obtained to compare between the original methods and the proposed modification. The result was shown in Table 1. Table 1 The error rate for difference normalization methods [42]. Normalization Method NO NORMALIZATION NORM GIC HISTEQ HMATCH LHE 5 × 5 LHM 5 × 5 LNORM 5 × 5

Error rate (%) 70.86 53.71 48.90 48.00 43.02 4.89 3.54 2.69

A fuzzy filter in discrete cosine transforms (DCT) was studied [43]. This method involved two factors; contrast limiting adaptive histogram equalization (CLAHE) and histogram equalization (HE) to find the background normalization. Vu & Caplier suggested a combination of two non-linear functions and Gaussian filter in order to eliminate varying illumination on the whole images [44]. A lot of researcher focus on post-processing as a main part of the illumination normalization approach, however, An et al. suggested pre-processing techniques with consist a Self-Quotient Image (SQI) normalization and Lambertian lighting model to eliminate the contrast variation images under varying lighting condition [45]. A new approach known as Multi-Scale Dual-Tree Complex Wavelet Transform (DT-CWT) was reported by Hu [46]. The luminosity and contrast problem was normalized by determining the lowest resolution wavelet, illumination invariant extraction, and the histogram remapping. Mendonça et al. investigated a few methods to find the best method for illumination normalization between homomorphic filter, wavelet and Learning Vector Quantization (LVQ) and by using Principle Component Analysis (PCA) the result shows the wavelet method is the best method compared to other methods [47]. In 2010, Amri et al. published a paper that reviewed two types of contrast enhancement techniques, namely linear contrast and non-linear contrast [48]. The linear type represented by three methods such Max-Min contrast, Percentage contrast, and Piecewise contrast. Four methods of the non-linear technique were used is Histogram equalization, Adaptive histogram equalization, Homomorphic Filter, and Unsharp Mask. After evaluation, the Piecewise contrast, and the Homomorphic Filter technique produced the best result. Fig. 5 shows the non-linear techniques image. In conclusion, many methods had been proposed by researchers to solve the illumination and contrast variation problem.

38

Penerbit


Akademia Baru

(a)

(b)

©

(d)

(e)

(f)

Fig. 5. Nonlinear Techniques: (a) Original Image Contrast, (b) Histogram Equalization Contrast Enhancement, (c) Adaptive Histogram Equalization, (d) LPF Homomorphic Contrast Enhancement, (e) HPF Homomorphic Contrast Enhancement, and (f) Unsharp Mask Contrast Enhancement [48].

3. Conclusion This paper presents an inclusive review of contrast enhancement methods. The vast number of available references shows that contrast enhancement was very important, especially to improve the image quality. The review focuses on the retinal images application in order to increase the blood vessel segmentation result. However, a few selected methods also were discussed and compared. The main target of the literature review was to find and explore the benefits of image enhancement algorithms and also to find the shortcomings in existing algorithms and techniques. In conclusion, many researchers agreed that it is very difficult and impossible to construct a perfect mathematical algorithm to solve the illumination and contrast problem at the same time. Image enhancement is found to be one of the most important elements in vision applications because it has the ability to enhance the visibility of the images. This study was done to find the gaps in the existing research and possible solutions to overcome these gaps in the future. Acknowledgment This work was supported by Ministry of Higher Education Malaysia under the Fundamental Research Grant Scheme (9003-00517), Bumiputera Academic Training Scheme (SLAB) (890909035027), and Fellow Scheme from University Malaysia Perlis, Malaysia. References [1] [2] [3]

Parveen, R., M. Nabi, F. A. Memon, S. Zaman, and M. Ali. "A Review and Survey of Artificial Neural Network in Medical Science." Journal of Advanced Review in Scientific Research 3, no. 1 (2016): 8–17. Porle, R. R., N. S. Ruslan, N. M. Ghani, N. A. Arif, S. R. Ismail, N. Parimon, and M. Mamat. "A Survey of Filter Design for Audio Noise Reduction." Journal of Advanced Review in Scientific Research 12, no. 1 (2015): 26–44. Mustafa, Wan Azani, and Haniza Yazid. "Illumination and Contrast Correction Strategy using Bilateral Filtering and Binarization Comparison." Journal of Telecommunication, Electronic and Computer Engineering (JTEC) 8, no. 1

39


[4] [5]

[6]

[7] [8] [9]

[10] [11] [12]

[13] [14] [15] [16]

[17] [18]

[19] [20]

[21]

[22]

[23]

[24] [25] [26] [27]

Penerbit

Akademia Baru

(2016): 67-73. Mustafa, Wan Azani, and Haniza Yazid. "Background Correction using Average Filtering and Gradient Based Thresholding." Journal of Telecommunication, Electronic and Computer Engineering (JTEC) 8, no. 5 (2016): 81-88. Mustafa, Wan Azani B. Wan, Haniza Yazid, Sazali Bin Yaacob, and Shafriza Nisha Bin Basah. "Blood vessel extraction using morphological operation for diabetic retinopathy." In Region 10 Symposium, 2014 IEEE, pp. 208-212. IEEE, 2014. Zheng, Yuanjie, Brian Vanderbeek, Rui Xiao, Ebenezer Daniel, Dwight Stambolian, Maureen Maguire, Joan O'Brien, and James Gee. "Retrospective illumination correction of retinal fundus images from gradient distribution sparsity." In 2012 9th IEEE International Symposium on Biomedical Imaging (ISBI). 2012. Kubecka, Libor, Jiri Jan, and Radim Kolar. "Retrospective illumination correction of retinal images." Journal of Biomedical Imaging 2010 (2010): 1–10. Leahy, Conor, Andrew O’Brien, and Chris Dainty. "Illumination correction of retinal images using Laplace interpolation." Applied optics 51, no. 35 (2012): 8383-8389. Vlachos, Marios Dimitrios, and Evangelos Spyros Dermatas. "Non-uniform illumination correction in infrared images based on a modified fuzzy c-means algorithm." Journal of Biomedical Graphics and Computing 3, no. 1 (2013): 6–19. Foracchia, Marco, Enrico Grisan, and Alfredo Ruggeri. "Luminosity and contrast normalization in retinal images." Medical Image Analysis 9, no. 3 (2005): 179-190. Meng, Xianjing, Yilong Yin, Gongping Yang, and Xiaoming Xi. "Retinal identification based on an improved circular gabor filter and scale invariant feature transform." Sensors 13, no. 7 (2013): 9248-9266. Bhandari, A. K., A. Kumar, and G. K. Singh. "Improved knee transfer function and gamma correction based method for contrast and brightness enhancement of satellite image." AEU-International Journal of Electronics and Communications 69, no. 2 (2015): 579-589. Ruggeri, Alfredo, and Simone Pajaro. "Automatic recognition of cell layers in corneal confocal microscopy images." Computer methods and programs in biomedicine 68, no. 1 (2002): 25-35. Gonzalez, Rafael C., and Richard E. Woods. "Image processing." Digital image processing 3 (2008). Øien, Geir E., and Per Osnes. "Diabetic retinopathy: Automatic detection of early symptoms from retinal images." In Proc. Norwegian Signal Processing Sym. 1995. Wallis, Robert. "An approach to the space variant restoration and enhancement of images." In Proc. of symp. on current mathematical problems in image science, naval postgraduate school, Monterey CA, USA, November, pp. 329-340. 1976. Kolar, Radim, Jan Odstrcilik, Jiri Jan, and Vratislav Harabis. "Illumination correction and contrast equalization in colour fundus images." In Signal Processing Conference, 2011 19th European, pp. 298-302. IEEE, 2011. Grisan, Enrico, Alfredo Giani, Elena Ceseracciu, and Alfredo Ruggeri. "Model-based illumination correction in retinal images." In 3rd IEEE International Symposium on Biomedical Imaging: Nano to Macro, 2006., pp. 984-987. IEEE, 2006. Mora, André D., Pedro M. Vieira, Ayyakkannu Manivannan, and José M. Fonseca. "Automated drusen detection in retinal images using analytical modelling algorithms." Biomedical engineering online 10, no. 1 (2011): 1–16. Mustafa, Wan Azani, Haniza Yazid, and Sazali Bin Yaacob. "Illumination correction of retinal images using superimpose low pass and Gaussian filtering." In Biomedical Engineering (ICoBE), 2015 2nd International Conference on, pp. 1-4. IEEE, 2015. Mustafa, Wan Azani, Haniza Yazid, and Sazali Bin Yaacob. "A review: Comparison between different type of filtering methods on the contrast variation retinal images." In Control System, Computing and Engineering (ICCSCE), 2014 IEEE International Conference on, pp. 542-546. IEEE, 2014. Saleh, Sami Abdulla Mohsen, and Haidi Ibrahim. "Mathematical equations for homomorphic filtering in frequency domain: a literature survey." In Proceedings of the International Conference on Information and Knowledge Management, pp. 74-77. 2012. Agarwal, Tarun Kumar, Mayank Tiwari, and Subir Singh Lamba. "Modified histogram based contrast enhancement using homomorphic filtering for medical images." In Advance Computing Conference (IACC), 2014 IEEE International, pp. 964-968. IEEE, 2014. Grigoryan, Artyom M., Edward R. Dougherty, and Sos S. Agaian. "Optimal Wiener and homomorphic filtration: Review." Signal Processing 121 (2016): 111-138. Xiao, Limei, Ce Li, Zongze Wu, and Tian Wang. "An enhancement method for X-ray image via fuzzy noise removal and homomorphic filtering." Neurocomputing 195 (2016): 56-64. Kedzierski, Michal, and Damian Wierzbicki. "Methodology of improvement of radiometric quality of images acquired from low altitudes." Measurement 92, (2016): 70–78. Forsyth, D.A. "Homomorphic filter." International Journal Computer Vision 5 (2010): 5–36.

40


[28]

[29] [30] [31] [32] [33] [34]

[35]

[36] [37]

[38]

[39]

[40] [41] [42] [43] [44] [45] [46] [47]

[48]

Penerbit

Akademia Baru

Zhu, Juhua, Bede Liu, and Stuart C. Schwartz. "General illumination correction and its application to face normalization." In Acoustics, Speech, and Signal Processing, 2003. Proceedings (ICASSP'03). 2003 IEEE International Conference on, vol. 3, pp. III-133. IEEE, 2003. Fan, Chun-Nian, and Fu-Yan Zhang. "Homomorphic filtering based illumination normalization method for face recognition." Pattern Recognition Letters 32, no. 10 (2011): 1468-1479. Ardizzone, Edoardo, Roberto Pirrone, and Orazio Gambino. "Illumination correction on MR images." Journal of clinical monitoring and computing 20, no. 6 (2006): 391-398. Tao, Qian, and Raymond NJ Veldhuis. "A Study on Illumination Normalization for 2D Face Verification." In VISAPP (1), pp. 42-49. 2008. Guo, Yucong, Xingming Zhang, Huangyuan Zhan, and Jing Song. "A novel illumination normalization method for face recognition." In Advances in Biometric Person Authentication, pp. 23-30. Springer Berlin Heidelberg, 2005. Park, Young Kyung, Bu Cheon Min, and Joong Kyu Kim. "A new method of illumination normalization for robust face recognition." In Iberoamerican Congress on Pattern Recognition, pp. 38-47. Springer Berlin Heidelberg, 2006. Salah-ELDin, Ahmed, Khaled Nagaty, and Taha ELArif. "An Enhanced Histogram Matching Approach Using the Retinal Filter’s Compression Function for Illumination Normalization in Face Recognition." In International Conference Image Analysis and Recognition, pp. 873-883. Springer Berlin Heidelberg, 2008. Vishwakarma, Virendra P., Sujata Pandey, and M. N. Gupta. "A novel approach for face recognition using DCT coefficients re-scaling for illumination normalization." In Advanced Computing and Communications, 2007. ADCOM 2007. International Conference on, pp. 535-539. IEEE, 2007. Lian, Zhichao, Meng Joo Er, and Juekun Li. "A novel local illumination normalization approach for face recognition." In International Symposium on Neural Networks, pp. 350-355. Springer Berlin Heidelberg, 2011. Park, Young Kyung, and Joong Kyu Kim. "A new methodology of illumination estimation/normalization based on adaptive smoothing for robust face recognition." In 2007 IEEE International Conference on Image Processing, vol. 1, pp. I-149. IEEE, 2007. Cheng, Yu, Zhigang Jin, and Cunming Hao. "Illumination normalization based on different smoothing filters quotient image." In Intelligent Networks and Intelligent Systems (ICINIS), 2010 3rd International Conference on, pp. 28-31. IEEE, 2010. Cheng, Yu, Zhigang Jin, and Cunming Hao. "Illumination normalization based on 2D Gaussian illumination model." In 2010 3rd International Conference on Advanced Computer Theory and Engineering (ICACTE), vol. 3, pp. V3-451. IEEE, 2010. Wang, Biao, Weifeng Li, Wenming Yang, and Qingmin Liao. "Illumination normalization based on Weber's law with application to face recognition." IEEE Signal Processing Letters 18, no. 8 (2011): 462-465. Lee, Hana, and Jeongtae Kim. "Retrospective correction of nonuniform illumination on bi-level images." Optics express 17, no. 26 (2009): 23880-23893. Santamaría, Mauricio Villegas, and Roberto Paredes Palacios. "Comparison of illumination normalization methods for face recognition." in: Work. Biometrics Internet, (2005): 27–30. Vishwakarma Virendra, P. "Illumination normalization using fuzzy filter in DCT domain for face recognition." International Journal of Machine Learning and Cybernetics (2013) 1–18. Vu, Ngoc-Son, and Alice Caplier. "Illumination-robust face recognition using retina modeling." In 2009 16th IEEE International Conference on Image Processing (ICIP), pp. 3289-3292. IEEE, 2009. An, Gaoyun, Jiying Wu, and Qiuqi Ruan. "An illumination normalization model for face recognition under varied lighting conditions." Pattern Recognition Letters 31, no. 9 (2010): 1056-1067. Hu, Haifeng. "Multiscale illumination normalization for face recognition using dual-tree complex wavelet transform in logarithm domain." Computer vision and image understanding 115, no. 10 (2011): 1384-1394. Mendonça, Michelle M., Juliana G. Denipote, Ricardo AS Fernandes, and Maria Stela V. Paiva. "Illumination Normalization Methods for Face Recognition." In Proc. of the 20th Brazilian Symposium on Computer Graphics and Image Processing, SIBGRAPI. 2007. Al-amri, Salem Saleh, N. V. Kalyankar, and S. D. Khamitkar. "Linear and non-linear contrast enhancement image." International Journal of Computer Science and Network Security 10, no. 2 (2010): 139-143.

41