Astha Mahajan et al, Int.J.Computer Technology & Applications,Vol 6 (5),677-686 COMPARITIVE STUDY OF VARIOUS MULTI-FOCUS IMAGE FUSION … · Fig-2 Image Fusion using DWT Principal

COMPARITIVE STUDY OF VARIOUS MULTI-FOCUS IMAGE FUSION TECHNIQUES

Dr. Vijay Kumar Banga, Astha Mahajan

Dept. of Electronics and Communication Engineering Amritsar College of Engineering and Technology, Amritsar

Abstract The main objective of vision fusion is to merging information from multiple images of the same view in order to deliver only the valuable information. The discrete cosine transforms (DCT) based methods of vision fusion are more appropriate and time-saving in real-time systems using DCT based standards of motionless images. In this paper, a well-ordered technique for fusion of multi-focus images based on variance calculated in DCT domain is shown. The overall objective of this paper is to compare different image fusion techniques. The comparison has shown that the Alternating Current (AC) coefficients calculated in DCT domain has quite better results. This paper ends up with the suitable future directions to extend this work. Keywords Image Fusion, DCT, DWT, PCA, IBLPCA Introduction Image fusion is the technique of combining various source images from sensor network into a single one, which contains a more accurate description of scenes, more useful and appropriate for both visual illustration and further processing. In this technique, several images of a scene are captured with focus on different objects are combined such that all the objects will be in focus in resulting image. With the rapid advancement in technology, it is now possible to obtain information from multi source images to produce a high quality fused image with spatial and spectral information. In other words, it is a process of combining all significant and corresponding information from various images of same source or various sources into a particular composite image without loss of information. Image fusion is an efficient means of getting all valuable data from source images and enables comparison and analysis of multi sensor data having

related information. Several situations in image processing require high spatial and high resolution in a single image. However, image fusion does not imply multi-sensor sources. There can be single-sensor and multi-sensor image fusion. The image fusion technique allows the integration of different information sources. The fused image can have complementary spatial and spectral resolution characteristics. While using image fusion technique, some requirements must be considered: • The fusion algorithm should not be discarded any

information contained in source images. • The fusion algorithm should not introduce any

artifacts or inconsistencies that can distract or mislead a human observer.

• The fusion algorithm must be reliable, robust and have capability.

Image Fusion Techniques Image Fusion is worked on one of the three different levels- signal, feature and decision. Signal level image fusion is also known as pixel-level image fusion where many different input images signals are combined to form a single fused image signal. Feature level image fusion is also known as Object level image fusion which fuses feature and object labels and information that have been collected from individual input images. Decision or symbol level image fusion shows fusion of probabilistic decision information obtained by local decision makers operating on the results of feature level processing on image created from individual sensors. Discrete Cosine Transform (DCT) - Normally, the digital images are displaying on a screen immediately after they are captured. There are two types of representation for digital image: spatial domain or frequency domain. The spatial domain analyses through

Astha Mahajan et al, Int.J.Computer Technology & Applications,Vol 6 (5),677-686

IJCTA | Sept-Oct 2015 Available [email protected]

677

ISSN:2229-6093

individual eye and frequency domain analyses through spatial domain. Human vision can realize low and medium spatial domain images but images with high spatial domain cannot be realized easily. Discrete cosine transform can convert spatial domain to frequency domain image. DCT is one of the important transformations used in digital image processing. DCT based image fusion is time saving technique in real time system. It divides the image into fixed size blocks in order to decide which source image should be selected to constitute the final resulting image. DCT can be used to decorrelate X images. After process of decorrelation every transform coefficient can be encoded discretely without losing compression efficiency. DCT coefficients are calculated for each block and fusion conventions are applied to get fused DCT coefficients. DCT are important to application in engineering, science and images etc. DCT coefficients are computed for each image and fusion rule is applied in order to get fused DCT coefficients, thereafter, IDCT is applied on the fused coefficients to produce the fused final image/block.

Fig-1 Image Fusion using DCT

Let DCT coefficient for first image block be X1 and DCT coefficient for second image block be X2. Suppose the image block is of size N x N and Xf be the fused DCT coefficient. Here, all DCT coefficients from both image blocks are averaged to get fused DCT coefficients. Xf (k1,k2) = 0.5 X1 (k1,k2) + X2 (k1,k2) Where k1, k2 = 0,1,2, …., N-1. The Discrete Cosine Transform of an image is defined as; X(k1, k2) = α(k1) α(k2)

∑ ∑ 𝑥𝑥(𝑛𝑛1,𝑛𝑛2)𝑁𝑁2−1𝑛𝑛2=0

𝑁𝑁1−1𝑛𝑛1=0 cos(𝜋𝜋(2𝑛𝑛1+1)𝑘𝑘1

2𝑁𝑁1) cos(𝜋𝜋(2𝑛𝑛2+1)𝑘𝑘2

2𝑁𝑁2)

for 0 ≤ k1 ≤ N1-1 and 0 ≤ k2 ≤ N2-1 where,

α (k1) = �

1√𝑁𝑁1

, 𝑘𝑘1 = 0

� 2𝑁𝑁2

, 1 ≤ 𝑘𝑘1 ≤ 𝑁𝑁1 − 1�

α (k2) = �

1√𝑁𝑁2

, 𝑘𝑘2 = 0

� 2𝑁𝑁2

, 1 ≤ 𝑘𝑘2 ≤ 𝑁𝑁2 − 1�

k1 and k2 are discrete frequency variables Similarly, the 2D inverse discrete cosine transform is defined as, X (n1, n2) = α(k1) α(k2) ∑ ∑ α(𝑘𝑘1) α(𝑘𝑘2)𝑁𝑁2−1

𝑛𝑛2=0𝑁𝑁1−1𝑛𝑛1=0 𝑋𝑋(𝑘𝑘1, 𝑘𝑘2)cos (𝜋𝜋(2𝑛𝑛1+1)𝑘𝑘1

2𝑁𝑁1)

cos (𝜋𝜋(2𝑛𝑛2+1)𝑘𝑘22𝑁𝑁2

) for 0 ≤ k1 ≤ N1-1 and 0 ≤ k2 ≤ N2-1 Discrete Wavelet Transform (DWT)- Discrete Wavelet Transform (DWT) is computed by consecutive low pass and high pass filtering of digital images. The transformed coefficients are fused and fused image is created by applying inverse Discrete Waveform Transform. As the wavelets do not represent long edges well in fused results, multi focus image fusion is performed by combining both the wavelet and curvelet transform to improve the quality. DWT needs large number of convolution calculations and it consumes much time and memory resources which impede its application for resource constrained battery powered visual sensor nodes. Firstly, Discrete Wavelet Transform (DWT) is calculated on each of source images to produce a fusion decision map based on set of fusion rules. The fused wavelet coefficient can be made from the wavelet coefficients of input images according to the fusion decision map. Finally the fused image is obtained by calculated the inverse wavelet transform.

First Image

Second Image

Fused Rule

Fused Image

DCT

DCT

IDCT



678

ISSN:2229-6093

Fig-2 Image Fusion using DWT Principal Component Analysis Method (PCA) - Principal component analysis (PCA) is a method used to reduce dimensionality and data representation. PCA is decorrelation method for image fusion in statistical sense and preserves important principal components. Principal components are derived from convergence matrix and its diagonalization by finding its eigen vectors and eigen values. The normalized eigen vector and values act as the weighted values and are multiplied with each pixel of input images. Principal components are linear combinations of optimally weighted observed variables. In PCA, the number of components extracted is equal to number of observed variables being analyzed. However, in most PCA analysis, only the first few components has the largest possible variance, only these first few components are retained, interpreted and used for analysis. Let X be a d-dimensional random vector and assume it to have zero empirical mean. The projection matrix V would be such that Y =VTX with the following constraints, the

covariance of Y, i.e, cov(Y) is a diagonal and inverse of V is equivalent to its transpose (V· I =VT). Using Matrix Algebra, cov(Y) = E{YYT} (1) cov(Y) = E{(XVT) (VTX)T} (2) cov(Y) = E{(XVT) (VXT) } (3) cov(Y) = VT conv(X)V (4) MuItiplying both sides of equation (4) by V, we get, Vcov(Y) = VVT conv(X)V = conv (X)V (5) Substituting equation (4) into the equation (5) gives, [𝜆𝜆1𝑉𝑉1, 𝜆𝜆2 𝑉𝑉2,……………………. , 𝜆𝜆𝑑𝑑 𝑉𝑉𝑑𝑑 ] = [conv(X)𝑉𝑉1, conv(X)𝑉𝑉2, … … … … … … … … . conv(X) 𝑉𝑉𝑑𝑑 ] This could be written as;

𝜆𝜆𝑖𝑖 𝑉𝑉𝑖𝑖 = con(X) 𝑉𝑉𝑖𝑖 Where, i =1, 2,…... d and Vi is an eigen vector of cov(X )

Iterative Block Level Principal Components Averaging (IBLPCA)- Iterative block level principal components averaging (IBLPCA) also known as a novel pixel level fusion algorithm has been proposed for fusion of noise free and noise filtered MR brain images. It evaluates principal components by splitting images into image blocks. Sizes of image blocks are decided based on average mutual information (AMI) between fused image and source image. Then the average of principal components of all blocks is evaluated to have weight for fusion rule.

Ref. No.

Author Year Techniques Features Applications/Benefits Remarks

[17] Phamila, Y.,and R. Amutha

2014

DCT

Extremely simple and energy efficient, suitable for resource constrained battery powered sensors for energy efficient fusion and subsequent compression.

A very simple, fast and energy efficient DCT based multi-focus image fusion scheme.

Can validate this approach on a sensor network testbed.

[19] Vijyarajan R, S Muttun

2014 IBLPCA Evaluate principal components of blocks of source images. By varying the number of blocks in the source images, this algorithm is iteratively applied to get fusion results with maximum average mutual information.

Noise free and noise filtered MR images.

The problem of the uneven illuminate can be neglected.

[18]

Li, Mingjing, and Yubing Dong

2013 Pixel-Level Pixel level image fusion refers to the processing and synergistic combination of information gathered by various

It can help people to overcome some technical difficulties, help doctors to

some color artifacts can removed to increase the

First Image Fused

Coeff. Map

Fused Image Second

Image

Coeff. Map 1

Coeff. Map 2



679

ISSN:2229-6093

imaging sources to provide a better understanding of a scene. It is the direct fusion in the original data layer, so the amount of information retained most.

diagnose diseases and locate lesions, help worker to complete product patrol and intelligent robot control.

performance.

[3] Sruthy, S., Latha Parameswaran, and Ajeesh P. Sasi

2013 (Dual Tree Complex Wavelet Transform)DT-CWT

has a better visual quality than the base methods.

Preserve substle texture regions of brain in MRI images. Ringing effects are reduced and it can retain the edge details more clearly.

can be extended further to include more hybrid techniques so that the user has a flexibility to have a choice of fusion techniques. The fusion trend utilizing Framelets to explore methods that are highly suitable and applicable to medical images.

[6] Desale, Rajenda Pandit, and Sarita V. Verma

2013 PCA, DCT, DWT

DWT based techniques are favorable than PCA and DCT as they provides better results for image fusion. Two algorithms based on DWT are, pixel averaging & maximum pixel replacement approach is discussed.

DWT based fusion techniques provide us good quality fused images

Some better techniques can be used for images which require high quality and precision.

[14] Zhang, Huaxun, and Xu Cao

2013 Medical Image Fusion based on wavelet theory

3 Steps has been used - image processing, image registration and image fusion. Image processing get across multi resolution characteristics of wavelet to denoise, Image registration pass the wavelet analysis to gain biggish change point and receive image edge to achieve quick and nice superposable, Image fusion use disassumble image to different frequency subband to save all information to have a perfect fusion.

Simple calculation, fast superposition and perfect fusion

Problem of the uneven illuminate can be removed.

[15] Liu, Lixin, Hongyu Bian, and Guofeng Shao

2013 Image Fusion scheme based on the lifting scheme of wavelets

Provides more valuable information, details of the outstanding characteristics, and good visual impression, which makes the fused image more suitable for human visual observation. It Performs the best in trade-offs in image processing speed and the quality of fused image

Faster speed, require less memory, easier to implement

Color artifacts can be remover for better results

[16] Prakash, Om, Richa Srivastava, and Ashish Khare

2013 Biorthogonal wavelet transform (BWT)

In this, two properties wavelet symmetry and linear phase of BWT have been exploited for image fusion because they are capable to preserve edge information and hence reducing the distortions in the fused image.

Improves fusion quality by reducing loss of significant information available in images.

Color artifacts can be remover for better results

[11] Jing Tian, Li 2012 adaptive An adaptive wavelet-based multi-focus Removes degree of the Approach could be



680

ISSN:2229-6093

Chen

image fusion in wavelet domain

image fusion approach assesses the sharpness of the image using the spreading characteristic of the wavelet coefficients distribution, which is modeled by a locally adaptive Laplacian mixture model.

image’s blur. further extended to be applied in the redundant or complex wavelet domains

[4] Patil, Ujwala, and Uma Mudengudi

2011 Hierarchical PCA

A pixel wise image fusion is done in pyramid decomposition and region based fusion during the fusion of each levels of source image pyramids.

Used for multimodal images like MRI and CT scan, multisensor images like visible and IR range images and multifocus images.

Work can be done with colour images also.

[7] Lavanya, A., K. Vani, S. Sanjeevi and R. S. Kumar

2011 Combined HIS and PCA transformations

Wavelet combined HIS and PCA transformations to produces better results spatially and spectrally for lunar image data compared to other methods.

Used for remotely sensed lunar image data in order to extract features accurately.

Automatic extraction of ridges and rilles can be performed on fused image to validate the proposed fusion technique.

[8] Haghighat, Mohammad Bagher Akbari, Ali Aghagolzadeh, and Hadi Seyedarabi

2011 Independent Component Analysis (ICA)

An adaptive algorithm based on the ICA fusion framework that maximizes the sparsity of the fusion image in the frame domain.

can be easily used in real time applications

Most of the existing work has focused on gray scale images not much work is done for color images.

[5] Hongzhi Wang, Lu wei, Cai he

2010 Discrete Cosine Transform (DCT)

An efficient approach for fusion of multi-focus images based on variance calculated in DCT domain.

Image quality and complexity reduction especially in visual sensor networks.

Boundaries between focused and out-of-focus areas can be preserved by using other filtering techniques.

[13] Pei, Yijian, Huayu Zhou, Jiang Yu, and Guanghui Cai

2010 Discrete Wavelet Framework

Can efficiently synthesis the useful information of the each source image retrieved from the multi sensor.

It Evaluates the quality and used in medical image fusion

Further verification can be done for more types of images.

[10] Shutao Li, Bin Yang

2007 Combination of Wavelet and Curvelet Transform

Two methods Wavelet and Curvelet Transform are combined together. Image is decomposed using curvelet transform and the coefficients are fused using wavelet-based image fusion method. Finally, the fused image is reconstructed by performing the inverse curvelet transform.

Simpler, faster and less redundant

Can be studied further for various combined method and to design more effective combination algorithm

[12] Drajic, Dejan, and Nedeljko Cvejic

2007 Dual-Tree Complex Wavelet transform (DT-CWT)

Dual-Tree Complex Wavelet transform (DT-CWT) is used to fuse the selected regions which is able to suppress the noise in the input images in different modalities, reducing significantly the

Used to fuse the sequences of images obtained from multimodal surveillance cameras

Problem of the uneven illuminate can be removed.



681

ISSN:2229-6093

distortion in the fused image, both visually and objectively.

[2] He, Dong-Chen, Li Wang, and Massalabi Amani

2004 Principal Components Analysis (PCA), Intensity-Hue-Saturation Transformation (IHS), High Pass Filter (HPF) and Wavelet transformation (WT).

Combining a high resolution image with a low resolution image with or without any spectral relationship existing between these two images and Preserving the spectral aspect of the low resolution image while integrating the spatial information of the high resolution image.

Used for images with different spatial resolutions.

In scope of this paper is that Low resolution images must be fused with a high spatial resolution image and colors of the original images can be preserved.

[9] Tang, Jinshan

2004 Contrast measure based DCT

A contrast based image fusion technique get comparable visual quality to the wavelet transform based fusion technique but it is simpler and less computationally expensive than the wavelet transform based fusion technique.

Simpler and less computationally expensive

Color artifacts can be removed for better performance.

[1] Nikolov, S. G., D. R. Bull, C. N. Canagarajah, M. Halliwell, and P. N. T. Wells

1999 3-D Wavelet Transform

Fusion by hard and soft thresholding, composite fusion, fusion of the WT maxima graphs. It may be combined with other 3-D image processing algorithms working in the wavelet domain, such as ’smooth versus textured’ region segmentation.

smooth versus textured region, volume compression where only a small part of all wavelet coefficients are preserved, and volume rendering

The problem of the uneven illuminate can be neglected.

Literature Review Sruthy S, Parameswaran L, et al. (2013) [3] has discussed that the image fusion involves set of images taken from different modalities of same scene. These fusion techniques are important in diagnosing and treating cancer in medical fields. This paper focuses on the development of an image fusion method using Dual Tree Complex Wavelet Transform. The results show the proposed algorithm has a better visual quality than the base methods. Also the quality of the fused image has been evaluated using a set of quality metrics. Dong-Chen H, et al. (2004) [2] discussed the challenges in the fusion of images integrating both the spectral aspects of the low resolution images and the spatial aspects of the high resolution images. They presented a new and original method of fusion, capable of (1) Combining a high resolution image with a low resolution image with or without any spectral relationship existing between these two images; (2) Preserving the spectral

aspect of the low resolution image while integrating the spatial information of the high resolution image. They did a comparison of their proposed technique to existing technique reported in the literature to prove their new proposed method innovation and uniqueness. Nikolov, S.G, et al. [1] has proposed a new approach to 3-D image fusion using a 3-D separable wavelet transform. Author has introduced a new framework for 3-D image fusion using the wavelet transform, rather than to compare the results of the various fusion rules. Wavelet transform fusion diagrams have been introduced as a convenient tool to visually describe different image fusion schemes. The integration of 3-DWT image fusion in the broader framework of 3-DWT image processing and visualization. Patil, U et al. (2011) [4] has proposed image fusion algorithm using hierarchical PCA. Authors had designed image fusion algorithm by combining pyramid and PCA



682

ISSN:2229-6093

techniques and carryout the quality analysis of proposed fusion algorithm without reference image. As there is an increasing need for the quality analysis of the fusion algorithms as fusion algorithms are data set dependent. They demonstrated fusion using pyramid, wavelet and PCA fusion techniques and carry out performance analysis for these four fusion methods using different quality measures for variety of data sets and show that proposed image fusion using hierarchical PCA is better for the fusion of multimodal imaged. Visible inspection with quality parameters are used to arrive at a fusion results. Haghighat, M et al. (2010) [8] proposed an efficient approach for fusion of multi-focus images based on variance calculated in DCT domain. The experimental results on several images show the efficiency improvement of their method both in quality and complexity reduction in comparison with several recent proposed techniques. DCT based methods of image fusion are more suitable and time-saving in real-time systems using DCT based standards of still image or video. Desale, R.P, et al. (2013) [6] has proposed the Formulation, Process Flow Diagrams and algorithms of PCA (principal Component Analysis), DCT (Discrete Cosine Transform) and DWT (Discrete Wavelet Transform) based image fusion techniques. The author has discuss the DWT based technique is more favorable than the other conventional techniques. There is two algorithms based on DWT are proposed, these are, pixel averaging & maximum pixel replacement approach. Lavanya, A et al. (2011) [7] has worked for multi-sensory and suggested that the image fusion is process of combining relevant information from high spatial resolution image and high spectral resolution image. They proposed a new image fusion method based on wavelet combined HIS and PCA transformations for remotely sensed lunar image data in order to extract features accurately. Results indicate that the fused image shows good spatial fidelity and the spectral resolution of the fused product was preserved after image data fusion. Hongzhi Wang, Lu wei, Cai he et al.(2010) [5] has proposed trained Independent Component Analysis (ICA) for image fusion. Common fusion rules can be used in the ICA fusion framework with promising

results, but has the shortcoming of the same fuse rule for the whole image. The author has proposed an adaptive algorithm which based on the ICA fusion framework that maximizes the sparsity of the fusion image in the frame domain. The experiment results show its excellent performance in image fusion. Haghighat M, Aghagolzadeh A et al. (2011) [8] has proposed an effici ent approach for fusion of multi-focus images based on variance calculated in DCT domain is presented. It can be easily used in real-time applications. The experimental results verify the efficiency improvement of proposed method both in output quality and complexity reduction in comparison with several recent proposed techniques. Tang J et al. (2004) [9] has proposed a new image fusion technique based on a contrast measure defined in the DCT domain is presented. The performance of contrast measure based technique is analyzed and compared with other image fusion techniques. Experimental results show that there is no difference in visual quality between the fused image obtained by our algorithm and that obtained by a wavelet transform based image fusion technique. But because our algorithm is carried out in the DCT domain, it is time-saving and simpler when the fused image needs to be saved or transmitted in JPEG format or when the images to be fused were saved in JPEG format. Shutao L , Yang B et al. (2004) [10] has proposed a multifocus image fusion algorithm based on combination of wavelet and curvelet transform. Although the fused results obtained by wavelet or curvelet transform individually are encouraging, there is still large room for further improvement because wavelets do not represent long edges well while curvelets are challenged with small features. Each of the registered images is decomposed using curvelet transform firstly. Then the coefficients are fused using wavelet-based image fusion method. Finally, the fused image is reconstructed by performing the inverse curvelet transform. The experimental results on several images show that the combined fusion algorithm exhibits clear advantages over any individual transform alone.



683

ISSN:2229-6093

Tian J, Li Chen et al. (2012) [11] has proposed exploiting the spreading of the wavelet coefficients distribution to measure the degree of the image’s blur. Furthermore, the wavelet coefficients distribution is evaluated using a locally adaptive Laplacian mixture model. The proposed sharpness measure is then exploited to perform adaptive image fusion in wavelet domain. Extensive experiments are conducted using three sets of test images under three objective metrics to demonstrate the superior performance of the proposed approach. Drajic Dejan, et al.(2007) [12] has present a method of fusing of the sequences of images obtained from multimodal surveillance cameras and subject to distortions typical for visual sensor networks environment. This method uses the Structural Similarity Measure (SSIM) to measure a level of noise in regions of a received image in order to optimize the selection of regions in the fused image. The region-based image fusion algorithm using the Dual-Tree Complex Wavelet transform (DT-CWT) is used to fuse the selected regions. The performance of the proposed method was extensively tested for a number of multimodal surveillance image sequences and outperformed the state-of-the-art algorithms, increasing significantly the quality of the fused image, both visually and in terms of the Petrovic image fusion metric. Pei, Yijian, Huayu Zhou, Jiang Yu, and Guanghui Cai, et al. (2010) [13] have proposed an improved discrete wavelet framework based image fusion algorithm. The improvement is the careful consideration of the high frequency subband image region characteristic. The algorithms can efficiently synthesis the useful information of the each source image retrieved from the multi sensor. The multi focus image fusion experiment and medical image fusion experiment can verify that proposed algorithm has the effectiveness in the image fusion. The author studies the quality assessment of the image fusion, and summarize and quantitatively analysis the performance of algorithms. Zhang, Huaxun, and Xu Cao, et al. (14) [2013] has introduced a way of medical image fusion based on wavelet theory. Medical image fusion has three steps, image processing, image registration and image fusion. Image processing get across multi resolution

characteristics of wavelet to denoise, image registration pass the wavelet analysis to gain biggish change point and receive image edge to achieve quick and nice superposable, image fusion use disassumble image to different frequency subband to save all information to have a perfect fusion. Simulation experiment proved it has advantages of simply calculation, fast superposition and perfect fusion in medical image fusion. It is a direction of medical reacher and clinic iatrology. Liu, Lixin, Hongyu Bian, and Guofeng Shao et al. (2013) [15] has proposed an effective multi-focus image fusion scheme based on the lifting scheme of wavelets. The pending images are decomposed by using the wavelet lifting scheme into four subbands firstly: LL, LH, HL, HH, and then subband LH, HL, HH, are synthesized to obtain three directions of high-frequency details of the images. The local luminance contrast, which is represented by weighted region energies, is calculated by the Gaussian kernel based on the high-frequency details. Thus the energy-based image fusion rule is applied to get a binary map by choosing the maximum energy between images. Experimental results show that the proposed method has a significant improvement of the fused image with no blocking effect or artificial effect and performs the best in trade-offs in image processing speed and the quality of fused image comparing with conventional methods. Prakash, Om, Richa Srivastava, and Ashish Khare et al. (2013) [16] have proposed a pixel-level image fusion scheme using multire solution Biorthogonal wavelet transform (BWT). Wavelet coefficients at different decomposion levels are fused using absolute maximum fusion rule. Two important properties wavelet symmetry and linear phase of BWT have been exploited for image fusion because they are capable to preserve edge information and hence reducing the distortions in the fused image. The performance of the proposed method have been extensively tested on several pairs of multifocus and multimodal images both free from any noise and in presence of additive white Gaussian noise and compared visually and quantitatively against existing spatial domain methods. Experimental results show that the proposed method improves fusion quality by reducing loss of significant information available in individual images. Fusion factor, entropy and standard



684

ISSN:2229-6093

deviation are used as quantitative quality measures of the fused image. Phamila Y., Amutha R. et al. (2014) [17] has proposed a simple and an efficient technique of multi-focus image fusion specially designed for wireless visual sensor systems equipped with resource constrained, battery powered image sensors employed in surveillance, hazardous environment like battlefields etc. The multi-focus fusion images are based on higher valued Alternating current (AC) coefficients calculated in DCT domain. This method overcomes the computation and energy limitations of low power devices and measures the resultant energy savings. The experimental results show efficient improvement in quality and energy consumption as compared to other fusion techniques in DCT domain. Li, Mingjing, and Yubing Dong, et al. (2013) [18] has proposed an overview of the most widely used pixel-level image fusion algorithms and some comments about their relative strengths and weaknesses. Some performance measures practicable for pixel-level image fusion are also discussed and prospects of pixel-level image fusion are made. Vijyarajan R. et al (2014) [19] has proposed a novel pixel level fusion called Iterative block level principal component averaging fusion by dividing source images into smaller blocks, thus principal components are calculated for relevant block of source images. Average of principal components of all the blocks provide weights for fusion rule, thus importance is given to blocks of source images. Iterations are incorporated in the form of size of blocks of source images which gives fusion results with maximum average mutual information. This algorithm is experimented for the fusion of noise free medical images and noise filtered of the same. The experimental results for both the cases show that the algorithm performs well in terms of average mutual information and mean structural similarity index. Liu L, Bian H. et al. (2013) [17] has proposed an effective multi-focus image fusion scheme based on the lifting scheme of wavelets. The pending images are decomposed by using the wavelet lifting scheme into four subbands firstly: LL, LH, HL, HH, and then subband LH, HL, HH, are synthesized to obtain three directions of high-frequency details of the images. The local luminance contrast, which is represented by

weighted region energies, is calculated by the Gaussian kernel based on the high-frequency details. Thus the energy-based image fusion rule is applied to get a binary map by choosing the maximum energy between images. There is a significant improvement of the fused image with no blocking effect or artificial effect. The method performs the best in image processing speed and the quality of fused image comparing with other methods. Gaps in literature By conducting the review it has been found that the most of the existing literature has neglected at least one of the following: 1. Many of the presented methods are based upon

transformations therefore it results in some color artifacts which can decrease the performance of the transform based vision fusion methods.

2. In existing work of fusion, there is problem of the uneven illuminate has been neglected.

3. Most of the existing work has focused on gray scale images not much work is done for color images.

Conclusion and future directions Image fusion is the technique of combining various source images from sensor network into a single one, which contains a more accurate description of scenes, more useful and appropriate for both visual illustration and further processing. The review has shown that the still many improvements are required in existing techniques. In near future a new technique which will combine the higher valued AC coefficients calculated in DCT domain based fusion with illuminate normalization to decreases the color artifacts which will be introduced due to the transform domain method i.e. DCT. The fusion process may degrades the sharpness of the edges in the digital images so to overcome this problem trilateral filter will also be integrated with proposed algorithm to improve the results further. References [1] Nikolov, S. G., D. R. Bull, C. N. Canagarajah, M.

Halliwell, and P. N. T. Wells. "Image fusion using a 3-D wavelet transform." (1999): 235-239.

[2] He, Dong-Chen, Li Wang, and Massalabi Amani. "A new technique for multi-resolution image fusion." In Geoscience and Remote Sensing Symposium, 2004. IGARSS'04. Proceedings. 2004 IEEE International, vol. 7, pp. 4901-4904. IEEE, 2004.



685

ISSN:2229-6093

[3] Sruthy, S., Latha Parameswaran, and Ajeesh P. Sasi. "Image fusion technique using DT-CWT." In Automation, Computing, Communication, Control and Compressed Sensing (iMac4s), 2013 International Multi-Conference on, pp. 160-164. IEEE, 2013.

[4] Patil, Ujwala, and Uma Mudengudi. "Image fusion using hierarchical PCA." In Image Information Processing (ICIIP), 2011 International Conference on, pp. 1-6. IEEE, 2011.

[5] Hongzhi Wang, Lu wei, Cai he. " An adaptive image fusion algorithm based on leA ." International Conference on Computer, Mechatronics, Control and Electronic Engineering (CMCE) 2010.

[6] Desale, Rajenda Pandit, and Sarita V. Verma. "Study and analysis of PCA, DCT & DWT based image fusion techniques." In Signal Processing Image Processing & Pattern Recognition (ICSIPR), 2013 International Conference on, pp. 66-69. IEEE, 2013.

[7] Lavanya, A., K. Vani, S. Sanjeevi, and R. S. Kumar. "Image fusion of the multi-sensor lunar image data using wavelet combined transformation." In Recent Trends in Information Technology (ICRTIT), 2011 International Conference on, pp. 920-925. IEEE, 2011.

[8] Haghighat, Mohammad Bagher Akbari, Ali Aghagolzadeh, and Hadi Seyedarabi. "Multi-focus image fusion for visual sensor networks in DCT domain."Computers & Electrical Engineering 37, no. 5 (2011): 789-797.

[9] Tang, Jinshan. "A contrast based image fusion technique in the DCT domain."Digital Signal Processing 14, no. 3 (2004): 218-226.

[10] Li, Shutao, and Bin Yang. "Multifocus image fusion by combining curvelet and wavelet transform." Pattern Recognition Letters 29, no. 9 (2008): 1295-1301.

[11] Tian, Jing, and Li Chen. "Adaptive multi-focus image fusion using a wavelet-based statistical sharpness measure." Signal Processing 92, no. 9 (2012): 2137-2146.

[12] Drajic, Dejan, and Nedeljko Cvejic. "Adaptive fusion of multimodal surveillance image sequences in visual sensor networks." Consumer Electronics, IEEE Transactions on 53, no. 4 (2007): 1456-1462.

[13] Pei, Yijian, Huayu Zhou, Jiang Yu, and Guanghui Cai. "The improved wavelet transform based image

fusion algorithm and the quality assessment." In Image and Signal Processing (CISP), 2010 3rd International Congress on, vol. 1, pp. 219-223. IEEE, 2010.

[14] Zhang, Huaxun, and Xu Cao. "A Way of Image Fusion Based on Wavelet Transform." In Mobile Ad-hoc and Sensor Networks (MSN), 2013 IEEE Ninth International Conference on, pp. 498-501. IEEE, 2013.

[15] Liu, Lixin, Hongyu Bian, and Guofeng Shao. "An effective wavelet-based scheme for multi-focus image fusion." In Mechatronics and Automation (ICMA), 2013 IEEE International Conference on, pp. 1720-1725. IEEE, 2013.

[16] Prakash, Om, Richa Srivastava, and Ashish Khare. "Biorthogonal wavelet transform based image fusion using absolute maximum fusion rule." InInformation & Communication Technologies (ICT), 2013 IEEE Conference on, pp. 577-582. IEEE, 2013.

[17] Phamila, Y., and R. Amutha. "Discrete Cosine Transform based fusion of multi-focus images for visual sensor networks." Signal Processing 95 (2014): 161-170.

[18] Li, Mingjing, and Yubing Dong. "Review on technology of pixel-level image fusion." In Measurement, Information and Control (ICMIC), 2013 International Conference on, vol. 1, pp. 341-344. IEEE, 2013.

[19] Vijayarajan, R., and S. Muttan. "Iterative block level principal component averaging medical image fusion." Optik-International Journal for Light and Electron Optics 125, no. 17 (2014): 4751-4757.



686

ISSN:2229-6093

Astha Mahajan et al, Int.J.Computer Technology & Applications,Vol 6 (5),677-686 COMPARITIVE STUDY OF VARIOUS MULTI-FOCUS IMAGE FUSION … · Fig-2 Image Fusion using DWT Principal

Documents