METHOD OF ENCODING CONTOURS WITH MUSICAL SOUND

Abstract

The article discusses technologies for converting visual information into sound form, which are actively developing at the present time. These technologies are used in auditory-visual systems for blind and visually impaired people. For sighted people, partial transfer of visual information to an acoustic signal will improve a general perception of information, facilitate a task of visual monitoring, and focus simultaneously on several visual fields. In the article it is proposed a method for constructing a sound image of an object's contour on a digital image using musical sound in a convenient for perception frequency range from 440 Hz to 1760 Hz. The contour defines the most important feature of the object is its shape and it is the most informative characteristic for recognition. In accordance with the proposed method, the sound image is formed on a basis of a onedimensional representation of the contour using the "angle-distance" signature is a function of the distance from centroid to the contour points with a uniform step on the angle. Centroid is analogous to the "point of view" that is a place where the eyes of a person are focused, when he begins to inspect an object bounded by the contour. When constructing a sequence of sound signals, the principle is used: a higher value of the signature corresponds to a higher frequency of the sound signal (higher note). Human hearing is characterized by a relatively high resolution to perceive changes in a musical sound frequency. The musical image constructed by this method carries visual information about the location of the object's contour points, while the task of recognizing the shape of the object is performed by the human brain. A total playback time of the sound image depends on the selected step on angle and the specified duration of sounding each signal in the sequence. Taking into account the inertia of hearing, it is recommended to set the duration of a signal 65 ms. Based on sound images, it is easy to interpret simple geometric shapes (circle, square, etc.), which are fundamental building blocks of more complex objects. Therefore, the recognition of their characteristic sound signature is an important step towards the interpretation of complex images.