# Introduction In this paper, a generic model to solve these multi-style [7] image TANV analysis problems has been proposed. The pair of book keeping aim to characterize the two domains, multi-scale [3] and semi-quad [6], the mapping functions is to reveal the relation between two variable variation's [4] [5] for noise variation analysis. The proposed model is called as auto-coupled image noise variation analysis and apply it to image noise variation analysis to validate its performance. The rest of the paper is organized as follows. Section 2 discusses about Multi-Scale Image Text Attribute Noise Variation (MSTANV) scheme. Analysis Model is presented in section 3. Section 4 presents the proposed model Multi-Scale Image Analysis method. Section 5 discuss the results and Section 6 concludes the paper. n many image recognition applications, people often send images from different sources and consequently they were received at different destinations. In addition, low resolution obtained at multiple receivers should be up-converted to a higher level of resolution for better interpretation at end user. Research works on such image analysis problems should benefit the practical applications under image interpretation and image human visual distinctive information analysis [1] [2]. # I The MSTANV scheme presented in this work adopts partitioned and relevant (P&R) 3LMW stretch (P&R3L) and two-stage decomposition structure is implemented. Here w H jx , w V jy and w D jz are the 3LMW particular constant quantity at horizontal, vertical and diagonal particular constant quantity. Let S xyz denote input image to analyse. Filters H jx , H jy and H jz used in P&R3L are replaced with (2 ? ixyz = ? xyz ? ixyz(5) where ? ixyz denotes a P&R operator, ? xyz =(? ? ixyz Txyz ? ixyz ) -xyz (? ? ixyz Txyz ? ixyz )(6) For mapping ? xyz , ? ixyz = ? xyz ? ixyz(7) Substituting ( 6) into (7), ? xyz =? xyz ? xyz =(? ? ixyz Txyz ? ixyz ) -xyz (? ? ixyz Txyz ? xyz ? ixyz ) (8) where ?? K k=xyz ? i?Ck:xyz |? xyz ? ixyz -? xyz ? kxyz | xyz xyz (11) resulting |? xyz ? ixyz -? xyz ? kxyz | xyz xyz = |? ixyz -? kxyz | xyz xyz . From these, by re-writing the (11) as, By substituting norm in (12), resulting in At reconstruction stage, the coefficients are analysed in to their original styles by the same dictionary mapping or book keeping. J(? k:yz,l:yz , ? k:yz , ? l:yz , ? k:yz,l:yz )= ? N k=1:yz ? M l=1:yz | Î?" P k:yz || Î?" Q l:yz |(? k,l:yz -Ä?" k,l:yz ) yz + ? N k:yz=1 ? k:yz (? M l=1:yz | Î?" Q l:yz |? k:yz,l:yz -yz)+ ? M l=1:yz ? l:yz (? N k=1:yz | Î?" P K:yz |? k:yz,l:yz -yz)-? N k:yz=1:yz ? M l=1:yz ? k:yz,l:yz ? k:yz,l:yz (15),andJ(? k:xz,l:xz , ? k:xz , ? l:xz , ? k:xz,l:xz )= ? N k=1:xz ? M l=1:xz | Î?" P k:xz || Î?" Q l:xz |(? k,l:xz -Ä?" k,l:xz ) xz + ? N k:xz=1 ? k:xz (? M l=1:xz | Î?" Q l: # V. Experiment Results and discussions As stated in above chapters, the TANV performance increases in variable variation's-information CI i:xyz of original signal S i:xyz and noisy quantity w i:xyz , related as CI i:xyz =I(S i:xyz +w i:xyz ±? :xyz ), but decreases in noise error criteria CN i:xyz . Therefore, good 3LMW basis for TANV should aim at maximizing CI i:xyz and minimizing CN i:xyz is implemented in this research work. Denoting P&R 3LMW as P&R3L (n :xyz ),where n=1,2,3,....,N and bi-P&R 3LMW is denoted by CDF(n :xyz ,n' :xyz ), where n is analytic 3LMW and n' is analyzed 3LMW. Proposed method has been implemented on nine 256 X 256 images Barbara, Boats, Butterfly, Cameraman, House, Straw, Lena, Baboon and Peppers as shown in fig 2, to compute their CI i:xyz and CN i:xyz values with respect to wavelets CDF(3,3) and P&R3L (4). In table 2 and table 3, listed the values of CI i:xyz and CN i:xyz when j :xyz =2 N-1 -1 and j :xyz = j :xyz =2 N-1 -2. These results represent the information of the first three 3LMW scales indication H, V and D as horizontal, vertical and diagonal subbands respectively. From the experimental results tabulated in table The PSNR results on a set of 9 images are reported in Table 5. From Table 5, clearly shows the proposed TANV method significantly outperforms for both uniform blurring and Gaussian blurring. # VI. # Conclusions In this paper, an image analysis algorithm has been introduced to improve the effectiveness of quality for images. In this paper, also presented a MSTANV scheme with a P&R3L 3LMW interscale model, which improved the signal estimation under noisy environment. Experimental results on image analysis demonstrated that the analysis approach can significantly outperform other leading image analysis methods. Finally, image analysis modelling techniques were employed to separate 3LMW particular constant quantity. The spatial classification of 3LMW pixels reduces the analysis estimation error and subsequently improving the TANV performance. # Global Journal of C omp uter S cience and T echnology Volume XV Issue II Version I Year ( ) ![jxyz = |L jxyz-xyz |? :xyz (1) where |L jxyz-1 | is the corresponding filter (|L D jx-1 |), (|L H jy-1 | and (|L V jz-1 |). The orientation vector are represented as, ? jxy (mx,ny)=[R jxy (mx,ny) R jxy+xy (mx,ny)] Txy =? jxy +? jxy ±?|L jxy-xy |? :xy (2) ? jyz (mx,ny)=[R jyz (mx,ny) R jyz+yz (mx,ny)] Tyz =? jyz +? jyz ±?|L jyz-yz ? :yz (3) ? jxz (mx,ny)=[R jxz (mx,ny) R jxz+xz (mx,ny)] Txz =? jxz +? jxz ±?|L jxz-xz |? :xz (4) where ? jxyz (mx,ny)=[ x jxyz (mx,ny) x jxyz+1 (mx,ny)] Txyz and ? jxyz =[ y jxyz (mx,ny) y jxyz+1 (mx,ny)] Txyz .](image-2.png "?") ![? xyz in to P xyz and Q xyz , as in to N xyz and M xyz respectively. Having a set of N xyz {Î?" P k:xyz ,1?k:xyz?N} for P xyz and a set of M xyz {Î?" Q l:xyz ,1?l:xyz?M} for Q xyz , where Î?" p k:xyz denotes the index of k in P xyz and Î?" Q l:xyz denotes the index of l in Q xyz . The QC Model corresponding to above projection is given by, J(? k:xy,l:xy , ? k:xy , ? l:xy , ? k:xy,l:xy )= ? N k=1:xy ? M l=1:xy | Î?" P k:xy || Î?" Q l:xy |(? k,l:xy -Ä?" k,l:xy ) xy + ? N k:xy=1 ? k:xy (? M l=1:xy | Î?" Q l:xy |? k:xy,l:xy -xy)+ ? M l=1:xy ? l:xy (? N k=1:xy | Î?" P K:xy |? k:xy,l:xy -xy)-? N k:xy=1:xy ? M l=1:xy ? k:xy,l:xy ? k:xy,l:xy (14),](image-3.png "") 1![Figure 1 : Flowchart of the proposed analysis based multi-scale image TANV analysis In the above fig 1, at coding stage, input images with styles were defined at initial. Dictionary or book keeping matrix is defined with the randomness in the image style selected, based on these coding is done. At tranform stage, coding coefficeints X are mapped in to relevnat coefficients Y image analysis values.At reconstruction stage, the coefficients are analysed in to their original styles by the same dictionary mapping or book keeping.](image-4.png "Figure 1 :") Analysis Modelj-xyz -xyz) zeros of the variable quantity of original filter H o:xyz ,so does for H o:xyz ||V jxyz . The analysed signal byproposed 3LMW, is an average of several MSTANVsignal by P&R3L. Noise variations of s j at scale j in adirection is (? xyz ,? xyz )=arg ?xyz,?k min [1/xyz|X xyz -? xyz ? xyz | xyz xyz + ? xyz 1 |? xyz | xyz ]+? xyz 2 ? Kk=xyz ? i?Ck:xyz |? xyz ? ixyz -? k | xyz xyz(10)where ? k stands for k xyz -th C k:xyz of particular constant quantity ? xyz . By rewriting the (10) as,(? xyz ,? xyz )=arg ?xyz,?k min [1/xyz|Xxyz -? xyz ? xyz | xyz xyz + ? xyz 1 |? xyz | xyz ]+? xyz 2 Text Attribute Noise Variation based Multi-Scale Image AnalysisIV.Multi-Scale Image Analysis Algorithm2015YearVolume XV Issue II Version I( ) Fxz |? k:xz,l:xz -xz)+ where ? k:xyz,l:xyz ,? k:xyz , ? l:xyz and ? k:xyz,l:xyz are the index constraints. In the scenario of image analysis, re-writing the (13), ? M l=1:xz ? l:xz (? N k=1:xz | Î?" P K:xz |? k:xz,l:xz -xz)-? N k:xz=1:xz ? M l=1:xz ? k:xz,l:xz ? k:xz,l:xz (16) we get by considering ? initial values, (? xyz ,? xyz )=arg ?xyz,?k min [1/xyz|X xyz -? xyz ?| xyz xyz + ? xyz 1 |? xyz | xyz ]+? xyz 2 ? K k=1: xyz ? xyz xyz ) + norm(? xyz n1 |? xyz | xyz ])+norm(? xyz n2 ? K k=xyz ? i?Ck:xyz |? ixyz -? kxyz | xyz xyz )Global Journal of C omp uter S cience and T echnologyi?Ck: xyz |? xyz ? i: xyz -? xyz ? k: xyz | 1: xyz (17) and ? y: xyz =arg ?: xyz min |? xyz | 1: xyz , by stating |S xyz -n xyz ? xy:yz:zx ? xyz | xyz