I. INTRODUCTION:

Filtering is perhaps the most fundamental operation of image processing and computer vision. In the broadest sense of the term “filtering,”[1] the value of the filtered image at a given location is a function of the values of the input image in a small neighbourhood of the same location. In particular, Gaussian low-pass filtering computes weighted average of pixel values in the neighbourhood, in which, the weights decrease with distance from the neighbourhood centre. Although formal and quantitative explanations of this weight fall-off can be given, the intuitions that images typically vary slowly over space, so near pixels are likely to have similar values, and it is therefore appropriate to average them together. The noise values that corrupt these nearby pixels are mutually less correlated than the signal values, so noise is averaged away while signal is preserved.

Bilateral filtering smoothes images while preserving edges, by means of a nonlinear combination of nearby image values. The method is non-iterative, local, and simple. It combines gray levels or colour based on both their geometric closeness and their photometric similarity, and prefers near values to distant values in both domain and range.

In contrast with filters that operate on the three bands of a colour image separately, a bilateral filter can enforce the perceptual metric underlying the CIE-Lab colour space, and smooth colours and preserve edges in a way that is tuned to human perception. Also, in contrast with standard filtering, bilateral filtering produces no phantom colors along edges in colour images, and reduces phantom colors where they appear in the original image

II. OVERVIEW:

We shall define low level feature to those basic feature that can be extracted automatically from an image without any shape information (information about spatial relationship).As such threshoulding is actually a form of low level feature extraction performed as a point operation. Naturally all of these approaches can be used in high level feature extraction, where we find shapes in images. It is well known that we can recognize people from caricaturist’s portraits. That is the low level feature that we encounter. It is edge detection and it aims to produce a line drawing. This are very basic techniques and more advanced once. The first order detector are equivalent to first order differentiation and naturally the second order edge detector operator are equivalent to one higher level of differentiation.

The first-order derivative is regularly used to express the gradient. The zero-crossing based methods search for zero crossings in a second-order derivative expression computed from the image in order to find edges, such as the Laplacian or a non-linear differential expression.

In point of conceptual view, the edge detection methods are categorized into contextual and non-contextual approaches. The non-contextual methods work autonomously without any prior knowledge about the scene and the edges. They are flexible in the sense that they are not limited to specific images. However, they are based on local processing focused on the area of neighbouring pixels. The contextual methods are guided by a priori knowledge about the edges or the scene. They perform accurately only in a precise context. It is clear that autonomous detectors are appropriate for general-purpose applications. However, contextual detectors are adapted to specific applications that always include images with same scenes or objects. Structurally, the edge detection methods incorporate three operations: differentiation, smoothing and labeling. Differentiation consists in evaluating the desired derivatives of the image. Smoothing lies in reducing noise and regularizing the numerical differentiation. Labeling involves localizing edges and increasing the signal-to-noise ratio (SNR) of the detected edges by suppressing false edges.

III. LITERATURE SURVEY:

A. Flow Based Image Abstraction

In this, an automatic technique was generated that generates a stylistic visual abstraction from a photograph. Our method is designed to convey both shapes and colours in an abstract but feature-preserving manner. First, it captures important shape boundaries in the scene and displays them with a set of smooth, coherent, and stylistic lines. Second, it abstracts the interior colours to remove unimportant details on the object surface while preserving and enhancing local shapes. What separates our approach from previous abstraction techniques is the use of a flow-based filtering framework. We employ existing filters for line extraction and region smoothing and adapt them to follow a highly anisotropic kernel that describes the “flow” of salient image features.

Figure 1.Low level image detection

We show that our approach improves the abstraction performance considerably in terms of feature enhancement and stylization, resulting in the production of a high-quality illustration from a photograph that effectively conveys important visual cues to the viewer. Such information reduction could facilitate quick data deciphering, as well as efficient data transmission over the network.

B. Edge and Line Feature Extraction based on Covariance Model

The purpose of this work is to build a feature extractor for Signals with multiple edges (or lines) with varying heights. Assuch, this work is an extension of Canny’s work and others. Since we don’t want to prejudice the design towards a filter, we prefer to use the term feature extractor rather than detection filter.

The starting point in the development of the feature extractor is a model in which the occurrence of edges in 1-D signals is described in terms of conditional auto covariance functions. Application of the Bayes criterion (minimum risk) with unit cost function for both the detection and the localization of the edges results in a feature extractor, the output of which can be interpreted as a sequence of log-likelihood ratios associated with the input signal.

C. Bilateral Filtering for Gray and Color Images

Bilateral filtering smooths images while preservingedges, by means of a nonlinear combination of nearbyimage values. The method is non-iterative, local, and simple. It combines gray levels or colors based on both their geometric closeness and their photometric similarity, and prefers near values to distant values in both domain and range. In contrast with filters that operate on the three bands of a color image separately, a bilateral filter can enforce the perceptual metric underlying the CIE-Lab colour space, and smooth colors and preserve edges in a way that is tuned to human perception. Also, in contrast with standard filtering, bilateral filtering produces no phantom colors along edges in color images, and reduces phantom colors where they appear in the original imageIn this paper, we propose a noniterative scheme for edgepreserving smoothing that is non-iterative and simple. Although we claims no correlation with neurophysiological observations, we point out that our scheme could be implemented by a single layer of neuron-like devices that perform their operation once per image. Furthermore, our scheme allows explicit enforcementof any desired notion of photometric distance. This is particularly important for filtering color images

IV. PROBLEM IDENTIFICATION:

Given an image that we view as a height field of pixel intensities, the task of image abstraction involves the following sub problems:

1. Line extraction. Capture and display “significant "height discontinuities.

2. Edge smoothing. Remove all “insignificant” height discontinuities.

3. Quality Analysis. Analyse the quality of the produced images.

Solving the first problem results in a “line drawing” while the second results in a “smoothed” or “flattened” height field. The combination of these two solutions often results in an image as shown in figures below.

Fig.2: Block diagram of Proposed Methodology

V. CONCLUSION:

To summarize, whatever I study I found that every model having some advantages and some limitation. This project is focussing on deriving the abstract and stylized interpretation of a photorealistic image into Non photo realistically rendered image. This will help us in rendering an image in such a manner that will provide the painterly look or cartooned abstraction on a variety of images featuring humans, animals, plants, buildings, still objects, and outdoor scenes.

VI. REFERENCES:

1. Henry Kang, Member, IEEE, Seungyong Lee, Member, IEEE, and Charles K. Chui, Fellow, IEEE “Flow-Based Image Abstraction” IEEE Transactions on Visualization and computer graphics, Vol. 15, NO. 1, January/February 2009

2. Kyprianidis, J. E. & Kang, H. (2011). Image and Video Abstraction by Coherence-Enhancing Filtering. Computer Graphics Forum, 30(2), pp. 593-602. (Proceedings Eurographics 2011)

3. C. Tomasi , Computer Science Department, Stanford University, R. Manduchi, Interactive Media Group, Apple Computer, Inc. ” Bilateral Filtering for Gray and Color Images” Proceedings of the 1998 IEEE International Conference on Computer Vision, Bombay, India

4. Jan Eric Kyprianidis, John Collomosse, Tinghuai Wang, and Tobias Isenberg,” State of the ‘Art’: A Taxonomy of Artistic Stylization Techniques for Images and Video”, IEEE Transactions on visualization and computer graphics, 2012 (Authors’ Version)

5. Jan Eric Kyprianidis Hasso-Plattner-Institute, Germany, Henry Kang University of Missouri, St. Louis, USA,” Image and Video Abstraction by Coherence-Enhancing Filtering”, Eurographics 2011 / M. Chen and O. Deussen(Guest Editors) Volume 30 (2011), Number 2

6. C. Tomasi R. Manduchi Computer Science Department Interactive Media Group Stanford University Apple Computer, Inc. Stanford, “Bilateral Filtering for Gray and Color Images” Proceedings of the 1998 IEEE International Conference on Computer Vision, Bombay, India

7. Ying Li, Tong Zhang, Daniel Tretter Imaging Systems Laboratory HP Laboratories Palo Alto, “An Overview of Video Abstraction Techniques”

Received on 29.04.2013 Accepted on 10.05.2013

Int. J. Tech. 3(1): Jan.-June. 2013; Page 39-41