INTRODUCTION

Multisensor image fusion is the method of conjoining significant evidence from two or more images addicted to a one image.The resultant image determined as an additional useful information than  at  all of the participated  images.In remote sensing solicitations,the accumulative obtainability of planetary accepted sensors provides a inspiration for various  image fusion procedures. Numerousconditions in image processing need great spatial and extraordinary spectral information in a particular image. Furthermost of the obtainableutensils is not proficient of  producing such informationinfluentially. Image fusion methods permit the combination of alteredevidencefoundations. The output image from fusion  may obligate harmonizing dimensional and supernatural information features. However, the average image fusion procedures can change the spectral evidence of the multispectral data while combining input images through fusion process.

[1] grants two elementary fusion areas, explicitly spatial domain and transform domain. Principal component analysis (PCA) which is dimensional province technique and discrete cosine transform (DCT), discrete wavelet transform (DWT), stationary wavelet transform (SWT), non-sub sampled contourlet transform (NSCT), and complex contourlet transform (CCT) which are transform domain procedures. Enactmentmeasures are executed to assess and authenticate the enactment of image fusion procedure. Investigationaloutcomesdirect that the image fusion techniquefounded on complex contourlet transform (CCT) is improved than formerapproaches.

[2] benevolences two methodologies for the image fusion, viz. spatial fusion and renovate fusion and nearby are practices for instance principal component analysis (PCA) which is dimensionl province and discrete wavelet transform (DWT), stationary wavelet transform (SWT) and discrete cosine transform (DCT) which are transmute domain procedures. Contrast  among PCA, DWT,  DCT with SWT is done.along with parameter like, spatial frequency (SF), standard derivation (SD), PSNR, NCC, etc.

[3] described proceduresfor image fusion using stationary wavelet transform (SWT) with the assistance of intuitionistic fuzzy set (IFS) processes. In IFS, fewer number of factors is actuallysuitable for computing membership importance. On the supplementary, SWT has shift invariance capabilityabove the DWT. Deliberated SWT founded image fusion procedures with numerous IFS processes for dissimilar data set and related the outcomes with separately.

[4]Numerousconventional multi-focus image fusion procedures and some unusual multi-focus image fusion proceduresarediscussed  and recommended the R-FCM (Rough-Fuzzy C-Means)method to improve the multi-focus image fusion approach in dynamic scene.

[5]projected  non subsampled contourlet transform based image fusion of multi-sensor satellite images. In the planned effort, interchange among the spectral bias and improvement of spatial evidence is observed while fusing input images. The harms of wavelet grounded fusion procedures for instance partial directivity, deficiency of phase evidence and shiftinvariant are spoken with the assistance of  Non subsampled contourlet transform. The Non subsampled contourlet benefits to recollect the inherent organizational evidence although disintegrating and reforming the image constituents. Result grounded instructions are used for constituent replacement for image fusion. The investigations are conceded out in contradiction of the existing state of art and perceived that the  projected scheme delivers auspicious outcomes pictorial and measureable. The proficiency of the projected scheme in the fused result is examined qualitatively by Isodata cataloging system.

[6]Satellite images are actuality castoff  in numerous domains of exploration completed the ages. Unique of the foremost restrictions of near by these images is their determination. Thus, image determinationimprovement is the primaryessentialstage in image processing. Image determinationimprovement is the method of transforming the images under deliberation so that gainedoutput image is additionallyproper than the input  image for essentialsolicitation. Image determination is imperativecharacteristic of any image. Improvedeminence image i.e. high-determination image productsimprovedoutcomes in image processing solicitation. Determinationimprovement can be done in numerousfields such as spatial and transform domains. Dissimilar transform domain approaches that are utilised for image determination improvement are viz., Discrete Wavelet transform (DWT), Stationary Wavelet Transform (SWT), Discrete Cosine Transform (DCT), Dual-Tree Complex Wavelet Transform (DT-CWT) etc. Beyond these, DT-CWT is initiate to be one of the maximumoperativeapproaches and also examined variousmethodsgrounded on exclamation for image determinationimprovement. The conversation and discrimination of various transform domain approaches for image determinationimprovement is conceded out on satellite standard images and investigational outcomes presenting the sovereigntyamongst these approaches for image resolution improvement is offered.

DWT BASED IMAGE FUSION

The input  images aredisintegrated into rows and columns by low-pass (L) and high-pass (H) sifting andsubsequent dejected sample at particular level to obtainan approximation (LL) and aspect (LH, HL and HH)quantities. Topping purpose is connected with smooth filters or low pass filters and wavelet function with high-pass filtering. Wavelet reconditions mark accessible an structure where an image is fragmented, through individual level following to a simpler purpose band[7].

The phases complicated in the image fusion determined  from wavelet transform based image fusion are given below.

•  Acquire source images to perform fusion.
• Relate the wavelet transform on source images doneselected wavelet at the anticipated level.
• Acquire the estimation and aspectquantities for input images.
• Combine quantities by anticipated image fusion instruction.
• Relate Inverse discrete wavelet transform on the compoundquantities and obtain the final fused image.

EXPERIMENTAL RESULTS COMPILATION AND DISCUSSIONS

IRS 1D satellite and  LISS-III sensor  images accomplished from NRSA Hyderabad, India are compiled  as input images for  image fusion to  obtainmore informative fused images. Fused images attained from different fusion methods, DWT, DCT variance, consistency verification with DCT variance and SWT are compared.Fused otcomes are evaluated by entropy, image quality index(IQI), spatial frequency (SF), peak signal to noise ratio (PSNR), standard deviation (SD), mutual information measure (MIM) between fused image and panchromatic, fused image and multi spectral image.

As SWT will not destroy constants at each transformation level SWT based fused image having higher values for entropy, IQI,PSNR, MIM and SD  indicates that SWT fusion outperforms DWT and DCT based fusion techniques.

The  overall results and discussions have proved  that proposed SWT based fusion techniques hav met the objectives of image fusion by improving spectral and spatial information as well

 CONCLUSIONS

Image fusion is a technique to converge MS and PAN images or many input images attained through different instruments into a one fused image incorporateshigh spectral and spatial validation. Fused images required in plenteousapplicationsviz image exploration, image indulgent, computer conception, cataloging, image appreciation, remote sensing, medical imaging, biometrics, video supervision. DWT, DCT and SWTimage fusion approaches are implemented here. Fused results are evaluated through varous assessment parameters. Because of the  basic property and potentiality of the SWT based fused results outperforms DWT and DCT based fused approaches. After SWT fused results DCT variance and CV with DCT variance based fused results improved fused image content.

ACKNOWLEDGEMENT