16-09-2016, 04:29 PM
1455003685-V5I60324.pdf (Size: 517.95 KB / Downloads: 7)
Abstract -Ancient paintings are the cultural heritage for ones country and must be preserved. With the passage of
time, painting gets damaged. The common deteriorations found in old paintings is Cracking. This paper discusses
several techniques for restoration of old painting which are suffering from cracks. The approach includes detection of
crack and removal of crack from digital painting. We can restore digital paintings using different image processing
techniques. Normally we use low pass filters to detect cracks because cracks are low illuminated areas in paintings.
But all low frequencies including brush strokes are passed by these filters, that’s why to identify cracks properly they
should be classified. Then we separate the thin brush strokes from the cracks by using method Semi-automatic crack
separation and Discrimination on the basis of hue and saturation which have been detected as crack. We restore the
image using local image information. For this purpose we use technique of order statistics filtering .Finally, using the
idea of median filtering filling technique the cracks are filled.
INTRODUCTION
Old painting which has great historic and artistic importance usually suffers from cracks or breaks[1]. These cracks can
be caused by aging, drying, and mechanical factors [2]. Age cracks are result from nonuniform contraction in the woodpanel
support or canvas of the painting, which stresses the layers of the painting. Drying cracks are caused by the
evaporation of volatile paint components and the consequent shrinkage of the paint. Finally, mechanical cracks usually
result from painting deformations due to external causes. The cracks in the paintings deteriorates the perceived image
quality. One can use digital image processing techniques to detect and eliminate the cracks on digitized paintings. This
“virtual” restoration can provide clues to art historians and the general public on how the painting was in its initial state,
i.e., without the cracks. It can also be used as a nondestructive tool for the planning of the actual restoration. Other
research areas which are closely related to crack removal is image inpainting which deals with the reconstruction of
missing or damaged image areas by filling in information from the neighboring areas, and disocclusion. Methods
developed in these areas assume that the regions where information has to be filled in are known. Methodology for
detection and restoration of cracks on digitized paintings, which adapts a number of image processing and analysis tools
is discussed.
The technique consists of the following stages:
i) crack detection ii) separation of dark brush strokes which have been misidentified as cracks
iii) crack filling (interpolation)
Crack classification is a prerequisite step for restoration of old digital paintings. User interaction, most notably in the
crack-detection stage, is required for optimal results
LITERATURE SURVEY
Ioannis Giakoumis [2] et al. introduced an integrated methodology for the detection and removal of cracks on digitized
paintings. The cracks were detected by thresholding the output of the morphological top-hat transform. then, the thin dark
brush strokes are removed using either a median radial basis function neural network on hue and saturation data or a
semi-automatic procedure based on region growing. Lastly ,crack filling using order statistics filters or controlled
anisotropic diffusion were performed.
Rousopoulos .et al. introduced a paper named “Determination of the method of drawing of prehistoric wall paintings via
original methods of pattern recognition and image analysis” [10]. In this paper a technique of construction of prehistoric
painting was shown. They proposed a algorithm that perform preprocessing of the boundary of the figures showing
within the painting, determines the patterns repetitions within the boundary of the represented components,
B. Cornelis et al [4] ,presented a new method for the virtual restoration of digitized paintings with special attention for
the Ghent Altarpiece , a large polyptych panel painting of which very few digital reproductions exist. They achieved their objective by detecting and digitally removing cracks. The detection of cracks were particularly difficult because of the
varying content features in different parts of the polyptych. Three new detection methods they proposed and combined in
order to detect cracks of different sizes as well as varying brightness. Semi-supervised clustering based post-processing
were used to remove objects falsely labelled as cracks. For subsequent inpainting stage, a patch-based technique were
applied to handle the noisy nature of the images and to increase the performance for crack removal. They demonstrated
the usefulness of their method by means of a case study where the goal is to improve readability of the depiction of text
in a book, present in one of the panels, in order to assist paleographers in its deciphering.
Sukhjeet Kaur [9] et al., developed a new algorithm that is nearest neighbour algorithm that can serve both the tasks of
detecting and removing the cracks, so the quality of the wall painting images can be improved. For better improvement
in the quality of digital wall painting ,another deformity is considered that is white spots which are detected as well as
removed .The nearest neighbour algorithm is improved by increasing the contrast and saturation. This algorithm provide
the more accurate result as compared to SIHF algorithm based on parameters that are Peak signal to noise ratio (PSNR)
and mean squared error (MSE). This algorithm gives the more accurate results than the SIHF algorithm as it remove the
more number of cracks and white spots.
III. CRACK DETECTION
Cracks can be categorized into two classes, bright cracks on a dark background or dark cracks on a bright background
[4]. Different crack detection techniques are simple thresholding, line detectors and various morphological filters.
Thresholding does not work well due to the noisy nature of the images and the presence of other crack like structures in
the image. The varying quality of the images and the difficulty of detecting cracks in low contrast zones requires several
pre processing steps. Different crack detection techniques that can be applied for the detection of both dark and bright
cracks has been introduced.
Cracks usually have low luminance and, thus, can be considered as local intensity minima with rather elongated
structural characteristics [3]. Therefore, a crack detector can be applied on the luminance component of an image and
should be able to identify such minima. A crack-detection procedure based on the top-hat .
CRACK FILLING
After identifying cracks and separating misclassified brush strokes, finally we have to restore the image using local
image information. For this purpose we use technique of order statistics filtering [6]. It is an efficient means to
interpolate the cracks and is to apply median or other order statistics filters in their neighborhood. All filters are applied
upon the cracks selectively.The core of the filter window passes through only the pixels of crack. If the filter window is
huge, the pixels of crack within the window will lie outside and will be rejected. Thus, the pixels of crack will be
assigned with the cost of one of the adjacent non crack pixels. A new filter known as a Modified Adaptive Median Filter
(MAMF) can be used which works on each RGB channel independently only on the crack pixel locations. So quality of
the content in other pixels is not affected .In addition to crack filling , this nonlinear filter, preserves the edges of the
paintings. The standard median filter could be used for filling the crack. Problem with this method lies on its fixed
window size .There could always be a possibility that crack pixel count in the local region may exceed the non crack
pixel count. This may result in replacing a crack pixel by another crack pixel.
We therefore propose a modified version of an adaptive median filter where the window size surrounding the crack pixel
can be varied.[6] This variation depends on the nature of pixels surrounding the crack pixels in the local region of
window. It runs only over the crack pixels so that information in other pixel is kept intact. The size of the filter window
surrounding each crack pixel is evaluated based on the number of crack pixels in the local region of the window. If the
number of crack pixels in the local region exceeds some threshold value, the size of the window is expanded till it falls
below the threshold. When the number crack pixels in the local region falls below this threshold level, the size of the
window satisfying the condition is treated as the order n of the filter for processing the crack pixel under observation. n
will be different for all crack pixels in the painting and is evaluated adaptively. Processing here refers to replacement of
crack pixel under observation by the median of the local observations.It is equal to one, among the neighboring pixels.
The filled crack pixels are defined by:
fi
=med (Ai-j
...Ai
..Ai+j) ...(3)
Where A are the pixels in the local region of the window and j = (n-l)/2 . For the color paintings the same process is used
on three independent channels individually and then combined them to obtain crack filled color paintings.