Automated recognition system for printed music

• US 5,706,363 A
• Filed: 04/07/1995
• Issued: 01/06/1998
• Est. Priority Date: 07/31/1991
• Status: Expired due to Fees

• Current Assignee: Yamaha Corporation
• Original Assignee: Yamaha Corporation
• Inventors: Kikuchi, Takeshi
• Primary Examiner(s): Boudreau, Leo
• Assistant Examiner(s): KELLEY, CHRISTOPHER S
• Application Number:
  US08/419,509
• Time in Patent Office: 1,005 Days
• Field of Search: 382/100, 382/112, 382/113, 382/182, 382/203, 84/76
• US Class Current: 382/113
• CPC Class Codes: G06V 30/10 Character recognition
• CPC Class Codes: G06V 30/155 Removing patterns interfering with the pattern to be recognised, such as ruled lines or underlines
• CPC Class Codes: G06V 30/18 Extraction of features or characteristics of the image
• CPC Class Codes: G06V 30/304 Music notations

Abstract

An automated recognition device for printed music which converts information of the printed music to two-dimensional image data and recognizes notes and signs from this data. Auto-correlation values along a vertical axis are obtained from pixels in a horizontal direction and a first peak is judged to be a line interval of staffs of the printed music. In a second aspect of the invention the number of hooks is determined on the basis of pixel number data existing in the horizontal direction in a region having a predetermined width on both sides of a note. In the third aspect of the invention, image data is removed from a start point by raster scan on the basis of consecutive data. In the fourth aspect of the invention, a shape of a sign in image data is projected on vertical and horizontal axes and the shape of the sign is assumed to exist in rectangular coordinates. The same processing is repeated and the shape of the sign is judged to exist in rectangular coordinates which can not be divided any further.

Claims

1. An automated recognition device for printed music comprising:

converting means for converting information of the printed music to two-dimensional image data;

line interval detection means for obtaining auto-correlation values in a direction of a vertical axis of distribution data of a number of pixels existing in a direction of a horizontal axis for each distribution data point along the vertical axis of the two dimensional image data and for detecting a first peak value of the obtained auto-correlation values as constituting a line interval of staffs of the printed music; and

line position detection means responsive to the line interval detection means for detecting positions of the staffs by adding data at five different points which are spaced apart by the detected line interval between adjacent points in the direction of the vertical axis of the distribution data, the data representing the number of pixels obtained in the direction of the vertical axis for each of the five different points and successfully shifting the five points along the vertical axis to thereby detect positions of the five points at a time when a maximum sum of the data is obtained, wherein the staffs of the printed music are detected by defining the positions of the five points as the positions of the staffs when the maximum sum of the data is obtained.

2. An automated recognition device for printed music comprising:

converting means for converting information of the printed music to two-dimensional image data;

pixel distribution detection means for obtaining distribution data in a direction of a vertical axis of a number of pixels existing in a direction of a horizontal axis in a rectangular region having a predetermined width on both sides of a stem of a note; and

hook number detection means for detecting at least one hook of the note on the basis of a length along the direction of the vertical axis of the distribution data corresponding to a number of times that the number of pixels along the horizontal axis consecutively exceed a predetermined value, the hook detection means being operative to detect the number of hooks by adding the length to a second predetermined value and by dividing this result by a third predetermined value, wherein recognition of the note is performed by employing the at least one detected hook.

3. An automated recognition device for printed music comprising:

conversion means for converting information of the printed music to two-dimensional image data;

first projection means for dividing out a region encompassing a sign in the direction of a first axis by projecting the image data with respect to the first axis;

second projection means for dividing out the region encompassing the sign in the direction of a second axis by projecting, with respect to the second axis, each image data output by said first projection means;

final projection judging means for comparing a result of a preceding dividing and a result of a current dividing by said first projection means or said second projection means and, when the compared results are equal to each other, determining the region encompassing the sign on the basis of the results of the division and, when the compared results are not equal to each other, causing a dividing operation by said first projection means or by said second projection means to be repeated; and

recognition means for recognizing the sign existing in the region determined by said final projection judging means.

4. An automated recognition device for printed music comprising:

a converting circuit that converts information of the printed music to two-dimensional image data;

a line interval detection circuit that obtains auto-correlation values in a direction of a vertical axis of distribution data of a number of pixels existing in a direction of a horizontal axis for each distribution data point along the vertical axis of the two dimensional image data and that detects a first peak value of the obtained auto-correlation values as constituting a line interval of staffs of the printed music; and

a line position detection circuit, responsive to the line interval detection circuit, that detects positions of the staffs by adding data at five different points which are spaced apart by the detected line interval between adjacent points in the direction of the vertical axis of the distribution data, the data representing the number of pixels obtained in the direction of the vertical axis for each of the five different points and successfully shifting the five points along the vertical axis to thereby detect positions of the five points at a time when a maximum sum of the data is obtained, wherein the staffs of the printed music are detected by defining the positions of the five points as the positions of the staffs when the maximum sum of the data is obtained.

5. An automated recognition device for printed music comprising:

a converting circuit that converts information of the printed music to two-dimensional image data;

a pixel distribution detection circuit that obtains distribution data in a direction of a vertical axis of a number of pixels existing in a direction of a horizontal axis in a rectangular region having a predetermined width on both sides of a stem of a note; and

a hook number detection circuit that detects at least one hook of the note on the basis of a length along the direction of the vertical axis of the distribution data corresponding to a number of times that the number of pixels along the horizontal axis consecutively exceed a predetermined value, the hook detection circuit being operative to detect the number of hooks by adding the length to a second predetermined value and by dividing this result by a third predetermined value, wherein recognition of the note is performed by employing the at least one detected hook.

6. An automated recognition device for printed music comprising:

a conversion circuit that converts information of the printed music to two-dimensional image data;

a first projection circuit that divides out a region encompassing a sign in the direction of a first axis by projecting the image data with respect to the first axis;

a second projection circuit that divides out the region encompassing the sign in the direction of a second axis by projecting, with respect to the second axis, each image data output by said first projection circuit;

a final projection judging circuit that compares a result of a preceding dividing and a result of a current dividing by said first projection circuit or the second projection circuit and, when the compared results are equal to each other, determining the region encompassing the sign on the basis of the results of the division and, when the compared results are not equal to each other, causing a dividing operation by said first projection circuit or by the second projection circuit to be repeated; and

a recognition circuit that recognizes the sign existing in the region determined by said final projection judging circuit.

7. A method of automated recognition of printed music, the method comprising the steps of:

converting information of the printed music to two-dimensional image data;

obtaining auto-correlation values in a direction of a vertical axis of distribution data of a number of pixels existing in a direction of a horizontal axis for each distribution data point along the vertical axis of the two dimensional image data;

detecting a first peak value of the obtained auto-correlation values as constituting a line interval of staffs of the printed music; and

detecting positions of the staffs by adding data at five different points which are spaced apart by the detected line interval between adjacent points in the direction of the vertical axis of the distribution data, the data representing the number of pixels obtained in the direction of the vertical axis for each of the five different points and successfully shifting the five points along the vertical axis to thereby detect positions of the five points at a time when a maximum sum of the data is obtained, wherein the staffs of the printed music are detected by defining the positions of the five points as the positions of the staffs when the maximum sum of the data is obtained.

8. A method of automated recognition of printed music, the method comprising the steps of:

converting information of the printed music to two-dimensional image data;

obtaining distribution data in a direction of a vertical axis of a number of pixels existing in a direction of a horizontal axis in a rectangular region having a predetermined width on both sides of a stem of a note; and

detecting at least one hook of the note on the basis of a length along the direction of the vertical axis of the distribution data corresponding to a number of times that the number of pixels along the horizontal axis consecutively exceed a predetermined value; and

detecting the number of hooks by adding the length to a second predetermined value and by dividing this result by a third predetermined value, wherein recognition of the note is performed by employing the at least one detected hook.