PATTERN RECOGNITION By Jean Constant From an original idea by Mikhail Bongard ABSTRACT The intent of this project is to explore M Bongard’s problems on visual pattern recognition, insert elements of color and texture in some selected examples and evaluate how esthetic components affect the scientific purpose of the initial demonstration. Keywords: pattern, pattern recognition, circle, triangle, plane, color, texture
1 Introduction Art and Science are two closely intertwined intuitive processes. Both disciplines feed and develop from each other in the ongoing process of exploring, explaining and asserting the physical world in terms that all can agree on. Mikhail Bongard is a Russian scientist who developed a rational to approach complex problems of visual pattern recognition. His research, compiled in a 1967 book called “Pattern recognition” contain numerous black and white line drawing examples from which he developed his observation (Fig.01). Artist theoricians, in particular V Kandinsky [3] and his Bauhaus colleagues, Ittten [4] and Albers [5] formulated their mapping of the creative environment based on a similar systematic approach of their field. How closely Science and Art can benefit from each other best practices is what the following will attempt to demonstrate
Figure 01: Bongard’s problem: circle above triangle / triangle above circle.
2 Methodology M Bongard original pattern outlines were filled with preselected color and texture schemes to complement the linear descriptive of the original line drawing. From an esthetic perspective the question was: - How does space affect the perception of color?
- How does color affect the perception of space? - Does Bongard’s demonstration hold if his sets of black and white outlines are rendered in complex schemes of color and texture, and if so, what does the viewer gain from the experience? To further appreciate this exercise it is important to be reminded of the specific premises from which M. Bongard operates and that he outlines as follow: “The purpose of pattern recognition is to discern patterns in the world. As patterns are sought, templates are made, unmade, and remade; slots are shifted from one level of generality to another; filtering and focusing are done; and so on. There are discoveries on all levels of complexity”.
3 Descriptive A Bongard problem-solving procedure has several stages, in which raw data gradually get converted into descriptions. The early stages are relatively inflexible, later stages become gradually more flexible. The final stages reach a property that Bongard calls tentatively, which means simply that the way a picture is represented is always tentative. The raw data are read in. Then they are preprocessed. This means that some salient features are detected. The name of these features constitutes a “vocabulary” of the problem; they are drawn from a general “salient feature vocabulary”. Some typical terms of the salient feature vocabulary are: line, segment, curve, horizontal, vertical, black, white, big, small, pointy, round, etc [1]. This descriptive catalogue matches in many respects the codification established by V. Kandinsky in “Points & Line to Plane” [3] as he records some fundamental elements of composition (point, line, plane) to map out the environment in which artists work and elaborate on these initial characteristics to formulate original statements. V. Kandinsky was very much aware of the mathematical nature of visual art and its value in helping comprehend better the mechanics by which a visual message is built and received by the viewer. Bongard postulates that in a second stage of preprocessing, some knowledge about elementary shapes is used; and if any are found, their names are also made available. Thus, terms such as: triangle, circle, square, indentation, right angle, arrow, etc... may be selected. This is roughly the point at which the conscious and the unconscious meet in humans. Tentative descriptions are made for one or a few of the visual schemes. They typically use simple descriptors such as: above, below, to the right of, to the left of, inside, outside. Also definite and indefinite numerical descriptors can be used: one, two, three, many, few, etc. More complicated descriptors may be built up, such as: further to the right of, less close to, almost parallel to, etc.. The challenge then, is to devise explicit rules that say how to: make tentative descriptions for each pattern; compare them with tentative descriptions for other patterns of either class, restructure the descriptions by adding, discarding or viewing the same information from another angle. Special agents called the sameness detectors are present on all levels of the program. They constantly run around within individual descriptions and within different descriptions, looking for descriptors or other things that are repeated. When some sameness is found, various restructuring operations can be triggered, either on the single description level or on the level of several descriptions [1]. The same criteria are often applied to describe an esthetic experience for the viewer’s benefit. Professional and art-critics alike analyze artistic statements in similar descriptive to illustrate and enhance the experience both at the scientific and at the emotional level. Light is the favored medium of the visual artist to enhance and increase the singularity of shapes and their positioning in relation to each other. Light is what shapes reliefs and creates forms. Depending on the surface it bounces back or is absorbed by the medium creating the outer definition of the object. In the
context of a two-dimensional surface, light is the medium used to provides the appearance of texture and create the dynamic of the objects and the composition. From pure white to complete darkness, I first processed the Bongard templates in a monochrome environment to focus both the singularity of each symbol and their association with each other within the composition (fig. 02).
Figure 02: Monochrome pattern. Color is defined by light. Its physical properties have been studied at length by scientists such as I. Newton who believed the eye was a passive recipient of light generating its own internal lighting and color effects, philosophers such as Goethe for which an important aspect of colors was to arouse emotions and more recently by art theoricians Itten and Albers among them, who centered their research on light and color as the most expressive mean of visual composition. To be consistent with the Bongard problems relating to pattern recognition, I added color to the plates at the last building stage both to increase specific patterns signals and to unite all the elements in one larger esthetic statement (fig.03).
Figure 03: Texture & color pattern
4 Conclusion Twenty-two plates were created following this method (fig 04). They were shown to different audiences in different settings. Although the project did not follow the scientific rigor expected in a contained environment, the many similarities that emerged in the process validated both Bongard’s methodology and the more humble object of this project focused in visual perception. The level of awareness in recognizing patterns was significantly raised. Ensuing discussions focused on the compositions themselves and confirmed that the work was received both as a reflection and a celebration of the on going symbiosis between the two disciplines in the field of cognitive science, an intriguing attempt to reconcile a very abstract, clinical two-dimensional procedure with larger perceptual esthetic concern that benefited both the experiment and the audience. It can only be hoped that further collaborative effort in the fields of Mathematics, Science and Art will continue to add to our collective knowledge base and help deepen the understanding of pure science and the perception of it.
Figure 04: Full composition – light – texture – color.
References: 1. Bongard, Mikhail Moiseevitch. Pattern Recognition. Rochelle Park, N.J.: Hayden Book Co., Spartan Books. (1970). Original publication: Problema Uznavaniya, Nauka Press, Moscow, 1967. 2. Hofstadter, Douglas R. Gödel, Escher, Bach. New York: Basic Books. (1979). 3. Kandinsky, Wassily. Point and Line to Plane. Dover Publications, New York. (1979) 4. Itten, Johannes The Elements of Color. John Wiley & Sons; 1 edition (January 31, 1970) 5. Albers, Jospeh. Interaction of Color. Yale University Press; Revised ed. (10-10, 1975) 6. Foundalis, Harris. Research on the Bongard problems. http://www.foundalis.com/ 7. Constant, Jean. Pattern Recognition lecture notes. National Center for Genome resources. Santa Fe, NM. 2000.
©2012 Hermay NM. For further information/contact:
[email protected] 
