WO2001082263A1 - Method for authenticating artwork - Google Patents

Abstract

A method for authenticating a subject painting comprising the steps of first scanning a subject painting using a scanner and thereby generating electronic image data of the scanned painting image. The electronic image data is then inputting into a computer for processing. An approxpriate software program or custom software and a mathematical model is then used to create an electronic signature of particular brush- strokes of the subject painting. The computer then electronically compares the electronic brushstroke signature of the subject painting with selected reference brushstroke signatures of paintings of confirmed authenticity so that the authenticity of the subject painting may be determined.

Description

METHOD FOR AUTHENTICATING ARTWORK

Cross-Reference to Related Applications

This application claims priority under 35 U.S.C. 119 based upon U.S. Provisional Application Serial No. 60/199,972 filed April 27, 2000, the entire disclosure of which is incorporated herein by reference.

Background of the Invention a) Field of the Invention:

This invention generally relates to a method of determining the authenticity of an obj ect, and more particularly, to a method of determining the authenticity of a subject painting by digitally analyzing the characteristics of the paint strokes used to create the subject painting and comparing the analyzed data with known paint-stroke characteristic data of the artist and/or of a painting of confirmed authenticity. b) Description of the Prior Art:

It has long been a problem in the art world to authenticate paintings or other works of art. Such authentication is often desired when a potentially valuable painting is put up for auction by an individual or when one must be appraised by an insurance company, for example, or perhaps by an estate court. Conventional methods used to authenticate a subject painting usually rely solely on the opinion and expertise of an art historian, specializing in the particular artist in question and/or the particular period, such as late French impressionism. Numerous elements are included in the evaluation by an expert. These elements are typically subjective determinations drawing on the expert's experience and other background. The art historian will study the painting's subject matter, style and/or painting technique used, and visually examine known detail located within the subject painting, such as the face detail of the painted subjects (i.e., how the artist painted a person's eyes). Analysis by the expert may also include study of how the paint is laid down on the canvas, such as whether the paint is thick or thin also including study of the brush stroke technique. Many paintings created during the seventeenth, eighteenth, and nineteenth centuries have been analyzed using various processes including the use of radar, x-ray, and bright light to reveal underlying detail otherwise hidden by outermost layers of paint so that a painting's "signature" can be learned. This information may be used to establish techniques employed by the artists of the different time periods, as well as the techniques used by any particular artist.

The analysis yields information which is invaluable during a restoration process of a painting, but is also helpful in the determination of the authenticity of a particular painting. For example, owing to the cost of art supplies during the eighteenth century, it was very common for "struggling" painters to paint over already painted canvases, and also to paint on canvases which already supported several charcoal renderings. This underlying artwork of a particular painting is revealed through the application of sound wave and/or X-rays and may be used to help authenticate the particular painting.

Also, the exact composition of the oil paint used to create a particular painting may offer invaluable information in authenticating a period piece of artwork.

For example, many of the oil paints used during the late eighteenth century included high amounts of lead and identifiable color pigments from various parts of the world depending on the color, such as the reddish brown pigment; burnt sienna, from East India. A paint made from ground pigment bound together with egg-yolk and water is called tempera and was the most common technique for the production of easel paintings until the late fifteen century. This information may be useful when trying to authenticate a painting by Jean Fouquet and Cimabue.

Watercolor was popular with English eighteenth century landscape painters such as John Robert Cozens and JMW Turner, and pastels become popular in portraiture in the eighteenth century and was revived in the nineteenth century with the work of the Impressionists, such as Edgar Degas and Liotard.

Many of today's paints are synthetic, such as acrylic paints and usually include small amounts of lead, if any. A spectrophotometer may be used to analyze the exact composition of the paint used in a subject painting to determine if the measured paint composition matches known paint compositions commonly used during the particular era and by the particular artist.

A process known as carbon dating may also be used to determine the approximate age of the materials used to create the painting, including the paint, the canvas, the charcoal, and the picture frame. Unfortunately, carbon dating has its limitations. In particular, carbon dating is not sufficiently accurate to be useful to determine the age of a particular painting, often yielding results in a range of ages, not a specific age.

Although scientific analysis of the oil paint and underlying artwork is helpful in detecting a forgery or proving the authenticity of a particular painting, it is not uncommon for forgerors, drawn by the high value of various famous easel artwork, to make the added effort to copy the exact composition of oil paint used, and to reproduce any underlying artwork known to be present of a particular painting, even mimic the style of particular artist's brush strokes. The end result is that the forgery paintings are becoming much more difficult to detect using conventional subjective techniques.

U.S. Patent No. 6,017,218 of Thomas J. Bright entitled: "Brush Mark Analysis Method for Painting Authentication", discloses a method to determine the signature of a particular painting by identifying certain characteristics of the brush used by the artist to spread the paint across the canvas. The content of US 6,017,218 in its entirety is hereby incorporated by reference. According to US 6,017,218, every painter's brush will generate a unique brush stroke which defines the brush's signature and can thereby be used to authenticate a particular painting by the painter. The method disclosed in US 6,017,218 is provided below: 1. Refer to a known painting by the artist in question. Pick a sample area with very legible, strong, clear, brush mark imprints and look for repeats. Longer brush marks are better as the features can be seen over more territory. An area of the painting seen in one color is desirable and lighter colors produce better images for use. 2. Produce an image of one brush mark of the selected group (known brush mark), which has a clear preferably straight length. This preferably starts with a photograph, preferably made under raking light to emphasize the ridges and rills. Print the image as nearly as possible at exact actual size. Then make a good photocopy of it. This will be the image used in the comparison. 3. Prepare the image to give edge access to the brush mark. This typically will involve cutting the image so that the desired brush mark attributes are at the edge of the cut.

4. Search the painting under investigation for repeated brush marks. Select one (study brush mark) for study. 5. Prepare an image of the study brush mark. This step can be skipped if the comparison is going to be made directly on the subject painting. But there is often an advantage to using an image of the subject painting in that the two images can be on thin material which facilitates the comparison step. Prepare an image of at least a portion of the painting in question which has the study brush mark selected for comparison. This preparation is done by producing a photocopy of a photograph as described above sometimes an X-ray will be useful. This should also be done at exact actual size so that the comparison is of images of the same scale. If other than actual size is used, such as a magnification, they must be the same.

6. Align the known brush marks at the edge of the known image to the selected brush mark on the study image.

7. Evaluate the closeness of match by identifying and counting the points of match. A minimum of 12 points of match are necessary in order to establish brush mark authenticity. It is preferable to count all the points of match over the entire brush mark even if more than 12 are found. While 12 is the minimum, more points of match increase the likelihood of acceptability of the conclusion. Although the authenticating method described in US 6,017,218 is interesting, it is replete with problems and inherent inaccuracies and will not be suitable for accurately determining the authenticity of a subject painting. First, the Inventor of US 6,017,218 argues that prior art authentication techniques rely on subjective comparisons by experts and that, for the most part, no scientific or objective techniques were being used. However, the methods outlined in US 6,017,218 remain subjective because two images of brush marks are placed side by side so that their "closeness" may be evaluated by a person. The accuracy of the method of US 6,017,218 relies on the person's evaluation of "closeness" of the brush strokes, which, in effect, is no more accurate than an expert examining the subject painting using a loop.

Regardless, the method described in US 6,017,218 is based on the belief that a paint brush will yield consistent and repeating markings through the paint being applied to a surface, as long as the brush is held in a consistent manner. Applicant contends that although a painter's style of brush stroke may be consistent for certain types of brush strokes, a brush's bristle characteristics will vary tremendously owing to a variety of parameters and conditions. Some of these include: the age of the brush, the quality of the bristles, the type of bristles, whether any bristles have fallen from their supportive ferule, whether any bristle has become bent, cut or otherwise damaged as the artist vigorously creates his/her masterpiece, the angle and applied pressure of the brush as it is used by the artist, the relative rotational displacement of the brush with respect to the canvas as the brush is held and used by the artist, the type of paint used, the color, and the thickness and viscosity of the paint (i.e., how much the particular paint has been thinned). It is not uncommon for paint to dry on a paint brush before a painting is complete, for the bristles of a paint brush to become misshapened during use and for many of the bristles to be fall from the handle ferule. Also, painters typically cut and shape the tips of their paint brushes according to the desired effect they are trying to achieve and also to remove old dried paint clogging the brush's tip. Painters often use a variety of different types of paint brushes all throughout a single painting. Different types of brushes, brush stroke directions, and brush stroke movements are often used within the same region of the painting and the lines of distinction of one brush will invariably be masked by and/or blended with the lines of distinction of a second or subsequent brush. Also, depending on the type and viscosity of the paint used, many of the bristle marks created by a paint brush will have been absorbed by the movement and flow of the paint as it cures. Furthermore, certain types of paintings, such as watercolor, waterpaint, gouache, pastels, Fresco, prints (silkscreen), Sfumato, and oil-stick are either applied to a canvas or wall without using a bristled brush or will not retain any markings of the brush's bristles (such as with watercolor paint).

Consequently, the marks (or lines of distinction) created by the bristles of a particular paint brush are not in themselves sufficiently consistent or identifiable to be relied upon in determining the authenticity of a painting.

Applicants have recognized a need to provide an accurate, consistent and reliable technique for determining the authenticity of a subject painting which overcomes the many deficiencies of the prior art.

Summary of the Invention

A method for authenticating a subject painting comprising the steps of first scanning a subject painting using a scanner and thereby generating electronic image data of the scanned painting image. The electronic image data is then inputting into a computer for processing. An appropriate software program or custom software and a mathematical model is then used to create an electronic signature of particular brush- strokes of the subject painting. The computer then electronically compares the electronic brushstroke signature of the subject painting with selected reference brushstroke signatures of paintings of confirmed authenticity so that the authenticity of the subject painting may be determined. Brief Description of the Drawings

Fig. 1 is an overall schematic of a scanning and computer system used to convert a subject painting into an electronic data file which is then manipulated according to the present invention; and Fig. 2 is an enlarged partial view of the carriage assembly of the scanning system shown in Fig. 1 , according to the invention.

Detailed Description of a Preferred Embodiment

Similar to the fact that every person has unique handwriting, every painter has a unique stroke signature, or several signatures depending on the type of painting and the subject matter being painted. The impressionistic painter Monet, for example, may have used one unique signature paint stroke when he painted the lilies in his famous painting "Water Lilies" and an entirely different signature painting stroke when he painted the sky portion of the same painting. The water lilies would entail quick short-stroke movements, whereas the clouds and sky detail would require broader sweeping paint strokes. Each of these subject regions of the painting would define a separate signature painting stroke.

In contrast to the above-described US 6,017,218 patent, it is the purpose of this invention to use mathematical models and image processing to identify, classify and store various signature painting strokes (not paintbrush marks) of each of many artists and to use the acquired information to establish the authenticity of a painting under study, or perhaps to reveal a painting as a forgery.

Referring to Figs. 1 and 2, a subject painting 10 (a painting under analysis) is positioned within a scanner 12 (the painting is laid flat - perpendicular to the surface of the paper - showing the edge of the painting). The scanner 12 may be a conventional planar- type scanner, including a carriage 14 supporting a light source 16, and a receiving sensor 18. As is well known, the carriage 14 is laterally displaced parallel and adjacent to the surface 22 of the subject painting 10 so that projected light 20 having a predetermined wavelength from the light source 16 reaches the painted surface 22. Any reflected light 24 is captured by the receiving sensor 18. As is understood by those skilled in the art, the receiving sensor 18 converts the received light into electrical signals which may be processed using a computer 26 and other appropriate signal electronics. As the carriage 14 moves the projected light 20 across the painted surface 22, the reflected light 24 will cause the receiving sensor 18 to generate electrical signals which convey the characteristics of the painting.

The light source 16 used in the scanner 12 may generate a variety of energy wavelengths including those in the visible spectrum, as well as ultraviolet and infrared. It is also contemplated that X-rays and/or sound waves may be directed at the painted surface 22 as well. Also, a combination of wave energies may be simultaneously or subsequently applied to the painted surface 22 to help extract the subtle characteristics of the painter's brush strokes and convert the information into an electronic digital signal. In addition, the projected light may be directed at the painted surface 22 at a prescribed angle 28 to exaggerate and reveal textural information of the paint used in the painting. Multiple scan passes may be performed with the projected light at the same or a variety of projection angles 26 and the extracted image data may be electronically combined. Regardless of the scanning methods used, the electronic digital signal generated by the scanning process will be manipulated and analyzed using image-processing techniques, according to the invention.

Most image-processing systems are based upon digital electronic technology that uses serial or one dimensional processing. An image is captured using a video camera or a scanner and is stored as a matrix of electrical signals. The image is then pre-processed to enhance edges, improve contrast, and otherwise isolate desired objects to be recognized. According to one embodiment of the present application, the desired objects to be detected are the various paint-stroke regions of a scanned subject painting.

As described in greater detail below, a comparison function compares an enhanced version of the image data (a signature) to one or more stored reference images. These processing and comparison functions are typically performed by standard microelectronic, digital equipment on a bit-by-bit or vector basis. According to a first aspect of the invention, the scanned image to be processed using at least a second order statistical mathematical model so that a mathematical representation of the texture of a sampled region or brush stroke may be generated. For second order statistical analysis, a two-point probability density function (2P- PDF) may be used as the model.

The purpose of this invention is to use an appropriate mathematical model to create a statistical signature of the texture of the scanned painting so that the generated or established signature may be electronically compared to other such signatures stored in a database and the authenticity (a match) of the subject painting may thereby be confirmed.

According to another aspect of the invention, the scanned image to be processed is converted into a Fourier or other transform. The Fourier transform presents information about the image of the object in a very useful, symmetrical pattern which represents the object in terms of its spatial frequencies. Since the calculation of a Fourier transform on a digital computer may take time, Fast Fourier

Transforms (FFT) will be computed and parallel processing techniques be used as well to further reduce the processing time. An image may also be optically processed to analyze it using Fourier transforms and, in this case, the processing occurs at the speed of light. It is likely that the Fourier transform in itself will not be sufficient to discriminate between textures, unless the Fourier transform is used in combination with statistical analysis, such as the full 2P-PDF.

Once generated, the acquired image may be processed by quantifying the light from a preselected number of spatial domains or segments of the acquired image. These quantities may then electronically processed to provide an abbreviated or composite characteristic signature of the acquired image, and thus of the object upon which it is based, or as in the present application, the textures and/or the spatial regions which make up each individual or selected groups of paint-brush strokes of a particular subject painting. In comparison to the time and expense involved when dealing with entire transform images, the signatures of the brush strokes of selected regions may be rapidly and economically obtained and then evaluated to determine whether the object does or does not conform to preselected standards.

Although possessing the above-noted benefits, an inspection system of the foregoing type must include signature-generating means, in addition to the other system components such as means for generating and storing electrical signal data representative of the image of the scanned subject painting, means for receiving such signal data and producing a visual image of the subject painting represented thereby, and means for producing a transform image of the subject painting from the visual image.

By producing one or more characteristic signatures derived from a known artist and/or a known painting of confirmed authenticity (and even known forged paintings), a database of signatures representing any of a variety of artists, paintings, or painting eras, including both authentic signature representations and known forged signature representations may be generated and enhanced over time so that all future subject paintings put under scrutiny by the present invention may be authenticated more accurately by merely comparing its measured brush stroke signatures with known brush stroke signatures stored in the database.

Through this technique a subject painting under study may objectively be analyzed and quickly compared with stored painting characteristics of the very same painting that the subject painting is purported to be and/or similar painting characteristics derived from previous analysis of paintings of a particular era or paintings of a particular artist.

The above-described database of painting stroke signatures will only be refined as more paintings are analyzed and their subsequent signatures stored. Much of this information may already be available by using the scanned data obtained during the restoration and/or study of a variety of paintings, by universities, museums, and art historians.

The data base can be created by any number of imaging techniques and stored in a digital storage system so that it can be reproduced by hard copy or by display on a screen and can be transmitted from place to place. Thus, the technique of comparing brush stroke signatures can be easily facilitated. The present method for authenticating a subject painting includes the following steps:

1. Scanning a subject painting using a scanner and thereby generating electronic image data of the scanned painting image; 2. Inputting the electronic image data to a computer for processing; 3. Using an appropriate software program or custom software and a mathematical model to create an electronic signature of particular brush- strokes of the subject painting; 4. Using the computer to electronically compare the electronic brushstroke signature of the subject painting with selected reference brushstroke signatures of paintings of confirmed authenticity; 5. Determining the authenticity of the subject painting in response to the outcome of the comparison step 4.

Although particular embodiments of the invention have been described and illustrated herein, it is recognized that modifications and variations may readily occur to those skilled in the art, and consequently it is intended that the claims be interpreted to cover such modifications and equivalents.

Claims

What is claimed is:

1. A method for determining the authentication of a subject painting, comprising the steps of: scanning a subject painting using a scanner and thereby generating electronic image data of the scanned painting image; inputting the electronic image data to a computer for processing; using an appropriate software program or custom software and a mathematical model to create an electronic signature of particular brush- strokes of the subject painting; using the computer to electronically compare the electronic brushstroke signature of the subject painting with selected reference brushstroke signatures of paintings of confirmed authenticity; determining the authenticity of the subj ect painting in response to the outcome of the comparison step 4.