FR2531783A1 - Method and device for the conversion of images obtained by sectorial scanning into line sweep images - Google Patents

Abstract

The device carries out real time conversion of original images into line sweep or field sweep images. It consists of a digital image memory 22 with sufficient capacity to store the samples taken over the N uniformly distributed scanning lines. An interface 24 for writing into the memory 22 is provided to take samples, on each scanning line n, at a fixed frequency for a given line chosen for each line in such a way that the samples from all the N scanning lines correspond to the same addresses in Y. An interface 30 for reading into the image memory comprises means for reading of the memory by lines each corresponding to an address Y. Means are provided for allocating to each pixel of address X the sample corresponding to the scanning line of order n given by:

Description

 Method and device for converting images obtained by sector scanning into line scan images.

 The present invention relates to the conversion, in real time, of images obtained by sector scanning according to N lines of shots in line presentation images. It finds a particularly important, although not exclusive, application in the conversion of images obtained by sectoral scanning into ultrasonic ultrasound.

 It is known that among the real-time ultrasound scanners used for medical diagnosis, those using a sector scan tend to take a prominent place. These devices use a mechanical system (rotating probes, oscillating), or an electronic scan. They provide at a high rate, typically of the order of 25 per second, images formed by the echoes collected by a number N, generally of the order of 100, firing lines also distributedangularly on an angular sector a.

 The simplest solution for visualizing the images thus obtained is to use an oscilloscope to which scanning ramps are applied in the X and Y directions varying as at.sinss and at.cos +, where <p is the angle of inclination of the firing line. But this mode of presentation has many disadvantages. The density of the lines varies a lot between the near area of the transducer and the far zone. Oscilloscopes giving a satisfactory gray scale, and thus making it possible to use echo quantification at many levels, are very expensive.

 These reasons lead to prefer the visualization on a television monitor, which implies a scan by lines ettrame iraster scan in English terminology) television standard.

 It would be a priori to think that the conversion to the television standard can easily be carried out by storing the samples taken along the N firing lines in a buffer memory which also makes it possible to freeze the image for a given sector scan if desired. Unfortunately, the direct transformation of polar coordinates into Cartesian coordinates at a very high rate (typically 2 MHz) requires a computation speed which is incompatible with the microprocessors currently available and therefore leads to the use of expensive and expensive calculators. the television standard representation of an image represented by pels whose density varies in large proportions, depending on whether one is in a region near or far from the transducer, leads to moiré effects in the region where the original image is at high spatial frequency. Finally, there are unaddressed points in the frame, which remain white and therefore give troublesome defects.

 The present invention aims to provide a method and a device operating in real time and for converting to the frame scan, and typically to television standard scanning, images obtained by sector scanning, avoiding the above defects and not only simple and relatively inexpensive means.

qui devisent, lorsque e = 900

For this purpose, the invention proposes in particular a real-time method of conversion to scanning by frames of images obtained by sector scanning following
N firing lines, characterized in that each successive firing line is taken from successive samples at a variable frequency according to the order n of the firing line so that they correspond to the same values of the address in Y for all the lines and stored in digital form, firing line by firing line, and an address transformation is performed on the stored samples to assign to each elementary point, or pel, address X of a television scan line of address Y the sample of order Y on the firing line n given by

who say, when e = 900

The invention also proposes an image conversion device obtained by sector scanning in line scan images comprising pixels regularly distributed in X and Y in a frame, comprising: a digital image memory of sufficient capacity to store the images; samples taken from the N uniformly distributed firing lines; 9 a memory write interface designed to take samples at a fixed frequency for a given line from each line of fire, chosen for each line so that the samples of all the lines shots correspond to the same addresses in Y; and a read-in-memory interface, comprising means for reading the memory by lines each corresponding to a Y address and means for allocating to each address pel X the sample corresponding to the given firing line n by

The invention will be better understood on reading the following description of a particular embodiment, given by way of non-limiting example. The description refers to the accompanying drawings in which
FIG. 1 is a diagram showing the distribution of the N-sector scan lines,
FIG. 2 is a simplified block diagram showing a possible constitution of the device according to the invention
FIG. 3 is a diagram showing the storage mode of the samples in the image memory.

It will be assumed that the invention is used to present, in real time, the echogram by sector scanning on an angle O along N firing lines equally distributed angularly. If the television scan presentation is performed on a matrix of M x M points, as shown in FIG. 1, the correspondence between the ADR addresses
X and ADR Y along X and Y and the polar coordinates of the sector scan is given, for a point P, by

 In these formulas - n is the order number of the firing line - t is the time - a is a constant proportionality coefficient.

The analog signal collected at each shot is subjected to a digital processing that must all at once - correct the excessive density of lines in the area
close to the transducer, - match a digitized sample to each
couple of address otherwise the points not addressed
would appear white.

 The method according to the invention uses a two-step coordinate conversion.

The calculation of the values of ADR Y is performed when writing to an image RAM organized in
M x M boxes each having a number of bits (bits) corresponding to the number of quantization levels chosen.

 The values of ADR X are calculated during playback and the values are each assigned to a pel when viewed on a television scanning screen.

 As a general rule, a quantization of at least four bits will be necessary and we will take M = 512.

 In Figure 2, the ultrasound device itself has been shown only schematically, since it will be of conventional constitution. It can be viewed as having a sector-scanning probe 6 associated with an echograph 7 controlled by a pilot card 8 provided with a time base 9. The ultrasound system and the probe may especially be of the type described in the patent application. FR 81 13445.

To store in an image memory 22 samples corresponding to evenly distributed values of ADR Y, the ultrasonic signal Z is sampled not at the constant frequency a, but at the frequency

<tb> pick <SEP> a
<tb><SEP> r
<tb> a <SEP> = <SEP> a <SEP> COS <SEP> e (n-1) <SEP> (3)
<tb><SEP> r <SEP> N <SEP> 2 <SEP> J
<Tb>
FIG. 2 shows a write interface 24 which makes it possible, for each firing line, to enter in a random access memory 22 samples corresponding to addresses
ADR Y determined. A n-order firing line will correspond to a writing column of the corresponding samples in the memory 22 (FIG. 3).

 The interface 24 comprises a read-only memory 10 (programmable read-only memory in general) in which are stored in advance the N values of the reduced sampling frequency given by the formula (3), the order number n of each firing line constituting the address where the corresponding value of ΔI ar is located. The stored digital value of the frequency ar is converted into analog voltage in a D / A converter 12 and applied to the control input of a voltage-programmed frequency oscillator 14 A counter 1Q causes, at the sampling frequency arç, the recording of the quantized amplitude of the signal Z in a buffer memory 18. This signal Z is digitized at the sampling instants by an analog / digital conversion circuit 20. four bits in the example given above) also controlled by the oscillator 14.

 The set of samples corresponding to an order firing line n is stored in the buffer memory 18, then transferred at a constant frequency into the main image memory 22 where it is assigned to the order n column of the memory .

 It should be noted that the buffer memory 18 must have a capacity corresponding to two firing lines. During the writing of the samples of a firing line, the preceding line is read and transferred to main memory 22.

 It should be noted that this writing method, firing line shooting line, leaves no empty box in the memory. For all the N firing lines, samples are stored in memory corresponding to all the values of ADR Y.

 Once the memory 22 is filled (after N shots), the reading is performed and must be accompanied by the transformation of the addresses according to X. This transformation must avoid the appearance of the defects mentioned above.

The reading will be performed horizontally, ie line by line, in the memory 22 which has N columns of samples. For each horizontal line, corresponding to the same address ADR Y, we have successive values of n. The transformation to perform

<tb> is <SEP> then <SEP> I <SEP> E <SEP> e <SEP>
<tb> ADR <SEP> X <SEP> = <SEP> M2 <SEP> AD <SEP> Y <SEP> ad <SEP>. <SEP>. <SEP> tsTe (II <SEP> 2 <SEP> 1)]} <SEP> (4)
<Tb>
This transformation makes it possible to assign a sample to each of the addresses following X, thus regularly spacing the samples over the entire surface of the rendered image, and filling any gaps that may exist between two successive firing lines, in the parts of the image farthest from the transducer, by copying the value for the adjacent firing line.

 But the calculation by direct application of this formula would take a long time, incompatible with the reading rate required for a standard television scan. Indeed, for M = 512, the base reading frequency f is 8 MHz.

 To avoid this problem of speed of calculation, one operates in two stages.

For each television scan line (line Y), the N values corresponding to the line Y are first transferred into a capacity buffer corresponding to one line at the frequency f.
- Then we read the memory at the same frequency, but by performing a transformation of addresses. For this, at each of the times t which correspond to an address in X, the sample is extracted from the buffer which is located at the address corresponding to the first order firing line.

We see that the M values of n, corresponding to
M different values of X, are independent of Y and are therefore the same for all television scan lines. They can therefore be stored in a transcoding ROM at M positions.

 it is necessary to multiply each address in its turn by Y. For that it is enough to sample the data which leave the buffer memory not at the frequency f, but at a frequency f 'proportional to f and to Y, well of tidying up samples in an output buffer, the capacity of which corresponds to the maximum number of dots per television scan line, ie to M.

 These operations can be performed using a read interface 30 of the kind shown in FIG.

In order to make it possible at the same time to write to the first buffer memory, the transformation of the addresses and the reading of the second buffer memory, the interface comprises two buffer lines, each comprising an input buffer memory 32. or 32a and an output memory 34 or 34a.

 The buffer write means 32 (or 32a) and the reading means of the output buffer memory 34 (or 34a) consist of two counters 36 and 38 which receive pulses at frequency f of the driver card 8. The means making it possible to present on the data output 40 of the memory 32 (or 32a) the firing data of order n given by the formula (5) comprise a counter 42 which also receives pulses at frequency f and attacks the addressing input of a programmable read-only memory 44.

The circuit for multiplying the addresses by Y must take into account the fact that the frequency f 'varies over a very wide range, practically from 0 to several
MHz and that the transitions at frequencies f and f 'must be synchronous so that there is a satisfactory transfer between the memories.

The circuit shown schematically in FIG. 2 fulfills these conditions by using the fact that the addresses
X2 that must be provided on the addressing input of the memory 34 are given by
X2 = E (f'.t) (6) or, if f is the base frequency and so if t = X / f
X2 = E (f'.X) = E (XY) (7)
f M what can still be written
x X2 = E (EY) (8) 2 1 M
This last function can easily be calculated at a fast rate. Indeed, it is a simple iterative addition synchronized with the frequency f. In the interface 30 shown in FIG. 2, the calculation is carried out by a circuit which comprises a programmable read-only memory 46 which applies, to a first input of an adder 48, the successive values of Y / M.This value is added to the successive values of X2 in the adder which supplies a register 50 whose clock input receives pulses at frequency F. The register drives the clock input of the output buffer 34 and the second input of the adder 48. The samples are thus stored in their natural order, with a translation by the memory 34. This is read back to the frequency f using the counter 38. The digital data is applied to a converter 52. The analog output signals are supported on the conventional video circuit 54 of a television monitor 56.

 It can be seen that the device which has just been described makes it possible to transform in real time the polar coordinates of the ultrasound into Cartesian coordinates for the purpose of viewing according to the conventional television standard.

The functions are fulfilled using only digital circuits, which eliminates the problems of instability and temperature drift. These circuits operate at clock frequencies at most equal to the basic sampling frequency of the television line. In the case where this frequency is 8 MHz, it can be seen that it is well below 10 MHz and that current TTL logic circuits in MOS can be used. It is also understood that the characteristics of the converter are determined solely by read-only memories that can be modified to change the shape of the display, of the type of probe, or even to switch from sectoral scanning to linear scanning (ultrasound B in the case of ultrasound ultrasound).

 The invention is obviously not limited to the particular embodiment which has been described by way of example and it goes without saying that the scope of this patent extends to any variant remaining within the scope of equivalences.

Claims (7)

 1. Real-time process of conversion to frame scanning of images obtained by sectorial scanning along N firing lines, characterized in that, on each firing line in turn, successive samples are taken at a variable frequency following order (n) of the firing line so that they correspond to the same values of the address in Y for all lines and stored in digital form, firing line by line of fire, and an address transformation is performed on the stored samples to assign to each elementary point, or pel, address X of a Y-address television scan line the Y-order sample on the firing line (n) given by

 2. Device for real-time conversion of images obtained by sector scanning in line scan images comprising pels regularly distributed in X and Y in a frame, characterized in that it comprises: a digital image memory ( 22) of sufficient capacity to store samples taken from N regularly distributed firing lines; a memory write interface (24) (22j for taking, on each firing line (n), samples at a fixed frequency for a given line, chosen for each line so that the samples of all N lines corresponding to the same addresses in Y, and an interface (30) for reading in image memory, comprising means for reading the memory by lines each corresponding to a Y address and means for assigning each address p X the sample corresponding to the order firing line n given by  3. Device according to claim 2, characterized in that the write interface comprises a read-only memory (10) in which are stored in advance the N values of the reduced sampling frequency ar connected to the frequency of sampling at the center of the beam by the relation

 means 112, 14, 16) connected to the output of the read-only memory (10) for sampling the image signal at the frequency ar and registering it in a buffer memory (18), and means for reading at a constant frequency for transferring the contents of the buffer memory (18) into the image memories (22).  4. Device according to claim 2 or 3, characterized in that the interface (30) for reading the image memory (22) comprises an input buffer memory (32 or 32 a) capacity corresponding to a line scanning means for transferring the N sampled values corresponding to the Y-order line of the image memory (22) to the input buffer (32), and means for replaying the input buffer ( 32) for extracting, for each address in X, the sample located at the address corresponding to the firing line of order n.  5. Device according to claim 4, characterized by means for sampling the data which leaves the input buffer (32) at a frequency Fs proportional to the constant read frequency of the input buffer (32) and to Y.  6. Device according to claim 5, since it is active in that said sampling means (46, 48, 50) are provided for storing the samples in an output buffer memory (34) whose capacity corresponds to the maximum number points per scan line.  7. Device according to claim 6, characterized in that the sampling means of the data which leave the input buffer and write buffer in the output buffer memory comprises a memory (46) addressable by Y, a circuit an adder whose inputs are connected to the addressable memory by Y (46) and to the output of a register (50) whose data input is connected to the output of the adder and the clock input receives pulses at the write frequency in the input buffer.