Integrated circuit device for orthogonal transformation of two-dimensional discrete data and operating method thereof

Claim 1

A circuit device for orthogonal transformation of two-dimensional discrete data arranged in a square matrix of n rows and n columns, n being a natural number, comprising: first storage means including a plurality of storage elements for storing individual elements of a coefficient matrix for orthogonal transformation in one to one correspondence therein, the coefficient matrix being a square matrix of n rows and n columns, second storage means including a plurality of storage elements for storing data arranged in a square matrix of n rows and n columns in the one to one correspondence basis therein, third storage means including a plurality of storage elements for storing therein a result of a multiplication between the data matrix stored in said second storage means and the coefficient matrix stored in said first storage means, the result of the multiplication providing a square matrix of n rows and n columns, said storage elements of said third storage means storing the result of the multiplication in the one to one correspondence basis therein, pointer means for producing information of a position of one of said storage elements included in said first storage means in which an element of the coefficient data in the i-th row and the k-th column is to be stored, i being an integer equal to or greater than 1 and equal to or smaller than n, k being an integer equal to or greater than 1 and equal to or smaller than n, information of a position of one of said storage elements included in said second storage means in which an element of said array data in the k-th row and the l-th column is to be stored, k being an integer greater than 1 and equal to or smaller than n, l being equal to or greater than 1 and equal to or smaller than n, an information of position of one of said storage elements included in said third storage elements in which an element of a matrix of a result of a multiplication in the i-th row and the l-th column is to be stored, multiplying means operable in response to position information produced from said pointer means for reading and multiplying information of those of said storage elements of said first and second storage means specified by said pointer means, adder means for reading information of ones of said storage elements of said third storage means specified by said pointer means, adding a result of the multiplication by said multiplying means to the data read out from the specified storage element of said third storage means, and storing a result of the addition into the specified storage element of said third storage means.

Claim 12

A method of orthogonally transforming a two-dimensional data array arranged in a square matrix of n rows and n columns with a coefficient matrix of n rows and n columns for orthogonal transformation, n being a natural number, comprising the steps of: writing the coefficient matrix for orthogonal transformation into a first storage block including a plurality of storage elements arranged in a square matrix of n rows and n columns and each element capable of making a shifting operation in the column direction and also in the row direction; writing the two-dimensional data array into a second storage block including a plurality of storage elements arranged in a square matrix of n rows and n columns and each element cable of making a shifting operation in the column direction and also in the row direction; resetting a pointer block including registers arranged in a square matrix of n rows and n columns and cable of making a shifting operation in the row direction such that only diagonal ones of the elements of the matrix of said pointer block may output an activating signal; reading and multiplying contents of said first and second storage means in response to an activating signal from said pointer block; reading contents of storage elements of a third storage block in response to the activating signal from said pointer block, adding the thus read-out contents to a result of the multiplication, and writing a result of such addition to the selected ones of said storage elements of said third storage block, said third storage block including a plurality of storage elements arranged in n rows and n columns; shifting contents of said storage elements of said first storage block by one step in the row or column direction and then shifting contents of said registers of said pointer block by one step in the row direction; reading and multiplying, in response to the activating signal from said pointer block, contents of the corresponding ones of said storage elements of the corresponding one of said first and second storage blocks; adding the result of such multiplication to contents of those of said storage elements of said third storage block selected in response to the activating signal from said pointer block and writing a result of such addition to the selected ones of said storage elements of said third storage block; repeating the multiplying and adding steps by n times, and stopping, after completion of such repetition, a shifting operation of said first storage block while a shifting operation is effected by one step in the column or row direction in said second storage block and the pointer output of said pointer block is shifted by one step in the row direction, one of said first and second storage means storage blocks being adapted to make a shifting operation in the column direction while the other makes a shifting operation in the row direction; repeating the multiplying and adding steps in response to an output of said pointer block; transferring, after completion of the matrix calculation as a result of execution of the calculating steps by n×n times, contents of said third storage block to said second storage block; and repeating the multiplying and adding steps, wherein said first and second storage blocks make shifting operations in the direction different from shifting operations in the preceding shifting operation.

Claim 13

A method of orthogonally transforming a two-dimensional data array arranged in a square matrix of n rows and n columns with a coefficient matrix of n rows and n columns for orthogonal transformation, n being a natural number, comprising the steps of: writing the coefficient matrix for orthogonal transformation into a first storage block including a plurality of storage elements arranged in a square matrix of n rows and n columns and each element capable of making a shifting operation in the column direction and also in the row direction; writing the two-dimensional data array into a second storage block including a plurality of storage elements arranged in a square matrix of n rows and n columns and each element cable of making a shifting operation in the column direction and also in the row direction; resetting a pointer block including registers arranged in a square matrix of n rows and n columns and cable of making a shifting operation in the row direction such that only diagonal ones of the elements of the matrix of said pointer block may output an activating signal; reading and multiplying contents of said first and second storage means in response to an activating signal from said pointer block; reading contents of storage elements of a third storage block in response to the activating signal from said pointer block, adding the thus read-out contents to a result of the multiplication, and writing a result of such addition to the selected ones of said storage elements of said third storage block, said third storage block including a plurality of storage elements arranged in n rows and n columns; shifting contents of said storage elements of said first storage block by one step in the row or column direction and then shifting contents of said registers of said pointer block by one step in the row direction; reading and multiplying, in response to the activating signal from said pointer block, contents of the corresponding ones of said storage elements of the corresponding one of said first and second storage blocks; adding the result of such multiplication to contents of those of said storage elements of said third storage block selected in response to the activating signal from said pointer block and writing a result of such addition to the selected ones of said storage elements of said third storage block; repeating the multiplying and adding steps by n times, and stopping, after completion of such repetition, a shifting operation of said first storage block while a shifting operation is effected by one step in the column or row direction in said second storage block and the pointer output of said pointer block is shifted by one step in the row direction, one of said first and second storage means storage blocks being adapted to make a shifting operation in the column direction while the other makes a shifting operation in the row direction; repeating the multiplying and adding steps in response to an output of said pointer block; transferring, after completion of the matrix calculation as a result of execution of the calculating steps by n×n times, contents of said third storage block to said second storage block; and repeating the multiplying and adding steps, wherein said first and second storage blocks make shifting operations in the direction different from shifting operations in the preceding shifting operation.