SU1644162A1 - Device for adaptive moving smoothing - Google Patents

Abstract

The invention relates to computing and can be used in digital signal processing systems. The purpose of the invention is to improve the accuracy. The device contains N 1 delay elements (N is the sample size), a generator of 2 clock pulses, a square average calculator block 3, a mean square calculator block, a linear regression coefficient calculator block 5, adders 6 and 12, a limitation block 7 , a dividing unit 8, a square root extraction unit 9, a subtraction unit 10 and a multiplication unit 11. The device allows improving the quality of the signal smoothing distorted by additive Gaussian noise with zero mean. 1 il. Second Ramp Out SA Memory Divider

Description

6L ZENEITA Zadeo ki

Calculate the approximation errors

The invention relates to computing and can be used in digital signal processing systems, for example, to reduce image noise.

The purpose of the invention is to improve accuracy.

The drawing shows a block diagram of the device.

The device for adaptive glide smoothing contains N delay elements 1 (N is the sample size), 2 clock pulses generator, block 3 for calculating the square of the average, block 4 for calculating the average square, block 5 for calculating the linear regression coefficient and, first , block 8 division, block 9 square root extraction, block 10 subtraction, block 11 multiplication, the second adder 12.

The device works as follows.

The samples (samples) of the input signal are sequentially in accordance with the frequency of the generator 2 clock pulses are fed to the input of the first delay element 1. Each of the delay elements delays one sample of the input signal sample sequence. In each device operation cycle, unit 3 recursively calculates the square of the local average value of a (k) over a symmetric neighborhood of N points with respect to g (k-l), including this count, where l (N-1) / 2, using the following relation:

a (k) ls (k), S (k) S (k-1) + g (k) -g (k-N).

where S (k) is the sum of N consecutive samples of the signal. By block 4, in each clock cycle, the local average value of squares of samples of the signal b (k) is calculated according to the following relation:

. b (k) -I T (k); T (k) T (k-1) + g2 (k) -g2 (k-N),

where T (k) is the sum of the squares of the values of N consecutive samples of the input signal.

The set of blocks 3-6 performs a calculation of the variance of the piecewise linear approximation based on the following relationship:

D (k) b (k - a2 (k) - c2 (k) / r, where r 1 (1 + 1) (2I-M) 2/3, a c (k) is calculated using the recurrence relation

c (k) c (k-1} -S (k} f g (k-NH + - j g (k)).

The values of a (k) and (k) / i (l + lX2l + 1) are, respectively, the free term and the slope coefficient in the straight line segment equation, which are determined by

least squares method. The value of c (k) is calculated in the kth cycle of the device in block 5 for calculating the linear regression coefficient. From the output of the adder 6, the calculated value of the local dispersion D (k) is fed to the input of the division unit 8. To another input of block 8 receives the value

2

noise variance about (). Block 8 calculates the coefficient a (f), i.e. implements function a

2 (l) a (i) / D (l). In case of exceeding the dispersion 2

this noise, about (i) local dispersion D (i), the value of a (I) at point i is set to

1 using constraint block 7 to satisfy condition 0. a (i) s1.

Thus, the output of the device, which is the output of the second adder 12, receives the value of the next sample f (i) g (i) + a (i) g (i) -g (i) of the result of adaptive smoothing.

Claims (1)

The invention The device for adaptive glide smoothing, containing N series-connected delay elements (N is the sample size), the square average calculator, the average square calculator, the linear regression coefficient calculator, the square root extraction block, the subtraction block, the first adder and the generator clock pulses, the output of which is connected to the inputs of the synchronization of the delay elements, blocks calculating the square of the average, calculating the average square, calculating the coefficient of the line regression, square root and the first adder data input device coupled to the input of the first of the delay element and with the first information inputs of the units for calculating the square of the mean, calculating the mean square and calculating the linear regression coefficient, the output of the M-th element handles connected the second the information inputs of the average square, average square and linear regression coefficient calculation blocks; the output of the square average calculator output is connected to the third information input of the linear regression coefficient calculator; the information output of the square average calculator is connected to the first information input of the first adder and information input of the square root extraction unit, the output of which is connected to the input of the decremented subtraction unit, the input of the subtracted which is connected to the output (N + 1) / 2-th delay element, information outputs of the linear regression coefficient and average calculation units the squares are connected respectively to the second and third information inputs of the first adder, characterized in that. that, in order to improve the accuracy, a dividing unit, a limiting unit, a multiplication unit and a second adder are inserted in it, the output of the first adder being connected to the input of the divider unit whose divisible input is an input for specifying the noise dispersion of the device, the output of the divider is connected to the first input of the multiplication unit, the second input of which is connected to the output of the subtracting unit, the output of the multiplication unit is connected to the first input of the second adder, the second input connected to the (N + 1) / 2-ro output of the delay element, the output of the second adder is the output of the device.