Different image scaling algorithms in graphics. A correlative analysis

CORRELATIVE ANALYSIS OF DIFFERENT IMAGE SCALING ALGORITHMS IN GRAPHICS

Abstract

This study paper demonstrate to find out the what is basically scaling in terms of image? What are the basic functions used in image scaling, also we take a look on different scaling algorithms and try to find out which algorithms is best in which context and what are the flaws and limitation of these algorithms. We also try to give out complimentary solution that what should be needed in order to make the efficiency of the scaling algorithms more strong.

Keywords: Computer Graphics, Image Scaling, Scaling Algorithms,

1 Introduction

In the world of graphics, image scaling is the procedure of resizing a computerized picture. Scaling is a non-trifling process that includes an exchange off between productivity, smoothness and sharpness. With bitmap representation, as the extent of a picture is decreased or extended, the pixels that frame the picture turn out to be progressively obvious, making the picture seem "soft" if pixels are averaged, or jagged if not. With vector graphics the trade-off may be in processing power for re-rendering the image, which may be noticeable as slow re-rendering with still graphics, or slower frame rate and frame skipping in computer animation.

illustration not visible in this excerpt

An image scaled with nearest-neighbor scaling (left) and 2*Sal scaling (right) Fig 1.1

Apart from fitting a smaller display area, image size is most commonly decreased (or subsampled or downsampled) in order to produce thumbnails. Enlarging an image (upsampling or interpolating) is generally common for making smaller imagery fit a bigger screen. In “zooming” a bitmap image, it is not possible to discover any more information in the image than already exists, and image quality inevitably suffers. However, there are several methods of increasing the number of pixels that an image contains, which evens out the appearance of the original pixels.

2 Summary

In this section now we will take a look on the scaling algorithms from past to present in order to understand that what are previous needs, how they was achieved and in this era what sort of scaling algorithms are working in the universe of computer graphics.

Study¹ shows that Nearest-Neighbor Interpolation is one of the most old and widely used algorithm in the scaling of image. Basically functionality of this algorithm is to replace each pixel with number of pixels of the similar color, This means that id the image is zoomed or magnified as in most cases if there is no scaling algorithm then the image pixels should be black out but if this algorithm will implemented then the pixels which are empty to the larger resulting image will be filled by nearest neighbor color. Suppose the neighbor has the color of white so neighbor 2 should be filled by white color when image becomes large

illustration not visible in this excerpt

Fig 1.2 NNI Image

illustration not visible in this excerpt

Fig 1.3 Original Image

From the fig1.2 as the image is enlarged, the pixels box which was empty due to the change of size was filled by nearest neighbor color. This algorithm however will not work much as when image becomes too much large it will causes raster image and hence image cannot be identified properly, and sometime when colors are too much in single pixel it will causes the mixture of different colors, which is wrong,

In this stream a new Bilinear interpolation Algorithm[[2]} which shows Bilinear interpolation sounds like 'bilerp' as a nickname is process of filtering the surrounding texels, to smooth out any rasters or blurrey zigzags occurring between pixels, and giving the screen a smoother look. Basically we can say that it is the enhanced power of NNI Algorithm.

illustration not visible in this excerpt

Fig 1.4 Bilinear

illustration not visible in this excerpt

Fig 1.3 Original Image

As we can see from fig 1.4 that is done by Bilinear interpolation Algorithm and hence it is quite better as compared to the Nearest-Neighbor Interpolation in some extend and in some cases. As the matter of fact is problem is still same that image is still blurry and raster too. On the other hand we also came to the point that ³ Bicubic Interpolation Algorithm are basically mixture of 1st two algorithms which are works on neighbor side but as also do a surroundings textures to make the image more clear and more easy visible

illustration not visible in this excerpt

Fig 1.5 Bicubic Fig

illustration not visible in this excerpt

1.3 Original Image

Figure 1.5 shows that image is looking like more clear and visible. As the matter of fact this algorithm is mixing up two other algorithms and hence it will takes time and its time complexity is more as compared the the 1st two algorithms but fact is the result of the image quality will be more nice as compared to the other 2 algorithms.

In this time a new algorithm⁴ Fourier-based interpolation was introduced. We can say that it was introduced to wipe out demerits of the Bicubic Interpolation Algorithm. In these algorithm basically we try to find out the frequency domain with zero components, here zero components will be considered as those pixels are not filled up just because of image enlargement. Hence this algorithm also try to find out the value and if if crosses the section area it will causes the filling of color which will be widely used as in overall whole image. As the matter of fact this is basically the 4th step in the world of scaling algorithm

illustration not visible in this excerpt

Fig 1.6 Fourier Based

illustration not visible in this excerpt

Fig 1.3 Original Image

Fig 1.6 shows how much it is looking more clear and visible and every time to calculate a frequency domain frequency and then it anslysis becomes hell of work for the CPU, thus this algorithm fails when we talk about the process management as it causes a lot of process for simple task.

One of the most wide used are [5[Edge-directed interpolation algorithms as the matter of fact is totally focus on the edges of the image. As image is enlarged the 1st thing will be done on edge to make it smooth and able to understand. Fact is this algorithm is best of edge detection section or forensic IT section where we have to concern with only shape but here in the universe of computer graphics the whole image should be be clear in such a way that it will be easy to understand that what a image represents. Edge- directed interpolation algorithms aim to preserve edges in the image after scaling, unlike other algorithms which can produce staircase artifacts around diagonal lines or curves.in it there is also concept of [6]Vectorization An entirely different approach is vector extraction or vectorization. Vectorization first creates a resolution independent vector representation of the graphic to be scaled. Then the resolution-independent version is rendered as a raster image at the desired resolution. This technique is used by Adobe Illustrator Live Trace, Inkscape, and several recent papers. Scalable Vector Graphics are well suited to simple geometric images, while photographs do not fare well with vectorization due to their complexity.

Conclusion

From start to end on this document we have seen that what are the different algorithms which are used for scaling algorithms, their advantages and their disadvantages here we came to the point till to end that Vectorization method is the best method. Although it is takes time but gives maximum output. Hence vertorization algorithms are always be best when we talk about excellent result but in terms of time management nearest neighbor will be good as compared to other, it is because of course it just has to filled empty pixel so just checked the neighbor's color and fill it. Remember never any algorithm will be best in all the case like time and accuracy but some are good in time some are good in time. Thus this journey of Scaling algorithm will continue ... Paper id:

3 References

[1] "Edge-Directed Interpolation". Retrieved 19 February 2016.
[2] Jump up ^ Xin Li; Michael T. Orchard. "NEW EDGE DIRECTED INTERPOLATION" (PDF). 2000 IEEE International Conference on Image Processing: 311.
[3] Jump up ^ Zhang, D.; Xiaolin Wu. "An Edge- Guided Image Interpolation Algorithm via Directional Filtering and Data Fusion" (PDF).
[4] Jump up ^ K.Sreedhar Reddy; Dr.K.Rama Linga Reddy (December 2013). "Enlargement of Image Based Upon Interpolation Techniques" (PDF). International Journal of Advanced Research in Computer and Communication Engineering 2 (12): 4631.
[5] Jump up ^ Dengwen Zhou; Xiaoliu Shen. "Image Zooming Using Directional Cubic Convolution Interpolation". Retrieved 13 September 2015. Jump up ^ Shaode Yu; Rongmao Li; Rui Zhang; Mou An; Shibin Wu; Yaoqin Xie. "Performance evaluation of edge-directed interpolation methods for noise-free images". Retrieved 13 September 2015..
[6] Johannes Kopf and Dani Lischinski (2011). "Depixelizing Pixel Art". ACM Transactions on Graphics (Proceedings of SIGGRAPH 2011) 30 (4): 99:1–99:8. doi:10.1145/2010324.1964994. Archived from the original on 2015-09-01. Retrieved 24 October 2012.

4 Biblography

1. "Eagle (idea)". Everything2. 2007-01-18.
2. "Gmane Loom". Retrieved 19 February 2016. 3. , Maxim. "hq3x Magnification Filter". Retrieved 2007-07-03.
4. Hunter K. "Filthy Pants: A Computer Blog". Retrieved 19 February 2016.
5. libretro. "common-shaders/hqx at master · libretro/common-shaders · GitHub". GitHub. Retrieved 19 February 2016.
6. Byuu. Release announcement Accessed 2011-08-14.
7. "xBR algorithm tutorial". Retrieved 19 February 2016.
8. libretro. "common-shaders/xbr at master · libretro/common-shaders · GitHub". GitHub. Retrieved 19 February 2016.
9. zenju. "xBRZ". SourceForge. Retrieved 19 February 2016.
10. "Super-xBR.pdf". Google Docs. Retrieved 19 February 2016.
11. libretro. "commonshaders/ xbr/shaders/super-xbr at master · libretro/common-shaders · GitHub". GitHub. Retrieved 19 February 2016.
12. http://pastebin.com/cbH8ZQQT
13. "RotSprite". Sonic Retro. Retrieved 19 February 2016.