The article begins by discussing the historical context of computer vision, giving a story of Marvin Minsky’s early attempts in the 1960s to solve the difficult subject of computer vision. It highlights the field’s continuing difficulty, particularly in high-level picture-processing tasks like pattern recognition. Given these difficulties, the text recognizes that low-level computer vision algorithms have been developed and become available to even young programmers.
The article goes on to discuss how digital video lacks fundamental conceptual information and requires computer vision to extract meaning from it. The author then introduces three essential methodologies for computer vision research: frame differencing, background subtraction, and brightness thresholding. He stresses the importance of developing physical conditions in combination with software development for good computer vision. The text points out the need to increase the contrast between objects of interest and their surroundings in the physical environment in order for algorithms like background removal and brightness thresholding to function properly.
The article also emphasizes the significance of the physical environment in computer vision, because algorithm performance is highly dependent on the quality of the incoming video scene. It goes on to predict that as technology advances and software tools improve, computer vision applications will become more common in a variety of areas, particularly media art education and interactive artworks.
I like how the text discusses how computer vision continues to impact the field of interactive media and the tremendous potential computer vision offers for interactive artwork. The artwork Messa di Voce was very interesting to me in this article. I think it was an incredibly exciting idea to have computer graphics (using computer vision) create real-time graphics to interpret the sonic quality of the singers’ voices. I only wish I could have seen this demonstration live!