JPS6222183A - Image processor - Google Patents

Abstract

PURPOSE:To improve the processing efficiency and throughput by sharing one buffer memory with two image processors. CONSTITUTION:With sending an initialization command from a CPU to each of sequencers 9A and 9B, an instruction address is sent to microprogram memories 8A and 8B, and microcode instruction is sent to both arithmetic units 10A and 10B, and a buffer memory 6A, and at an arithmetic unit 10, a larger one out of internal arithmetic parameters is set at a master image processor 7A side. With sending a process start command to the sequencer 9A, an another processor go command is sent to a slave image processor 7B, and process routines are simultaneously stared sharing the buffer memory 6A.

Description

DETAILED DESCRIPTION OF THE INVENTION [Technical Field of the Invention] The present invention relates to an image processing device in a radiographic system used for medical diagnosis.

[Technical foreground of invention]

An example of an X-ray photography system among the radiography systems will be described with reference to FIG.

This X-ray photography system includes an X-ray imaging device 1 having a phosphor plate that irradiates a subject with X-rays and accumulates the X-rays that have passed through the subject as energy for an X-ray image;
The phosphor plate is scanned with excitation light having a wavelength of 500 to 800 nm, and the energy stored in the phosphor plate is excited from its traps to emit light having a wavelength of 300 to 500 nm. An X-ray image reading device 2 configured to measure light with a photodetector (for example, a photomultiplier tube, a photodiode, etc.) set to receive only light in that wavelength range, and an X-ray image reading device 2 After reading the X-ray image, the output signal of the photodetector is nonlinearly amplified,
Furthermore, after converting into a digital C signal with an A/D converter, this signal is subjected to frequency emphasis processing and gradation processing as necessary, and the results are stored in a storage means such as an image memory. The image data stored in the device 3 and the image memory is sequentially read out, converted into analog signals by a D/A converter, and further amplified by an amplifier, and then inputted to a recording light source to convert the XvA image data into optical signals. an X-ray image reproducing device 4; and an X-ray image recording device 5, which inputs the optical signal through a lens system and irradiates it onto a recording material such as a photographic film to form an X-ray image on the recording material.
The X-ray image formed on the recording material is observed and used for diagnosing the subject.

A conventional example of the image processing device 3 in such an X-ray photography system is shown in FIG.

The image processing device 3 includes a buffer memory 6 connected to the image data and control bus and storing X-ray image data;
It is connected to the CPU bus and includes an image processor 7 that performs frequency emphasis processing and gradation processing on the X-ray image data stored in the buffer memory 6 in accordance with commands sent from the CPU via the CPU bus. .

Next, an algorithm for X-ray image processing by this image processing device 3 will be explained.

This algorithm is executed by arithmetic processing according to the following equation (1).

S=γ(SO+β(So-3ux))...(
11 Here, So is the original image signal read out from the phosphor plate of the X-ray image reading device 2, and SUS is the unsharp mask signal (the original image is blurred so that it contains only frequency components lower than extremely low frequency components). 3us = ΣS・
, ./N. 1゛1 Further, β is an emphasis coefficient for extremely low frequency components, T is a gradation coefficient, and N is a parameter (mask size).

In the above formula (11), (S, +β(S, -3us)
) is known as frequency emphasis processing,
Further, the calculation process of r (So + β(S6-3us)) is known for X-ray image data subjected to frequency emphasis processing. (Refer to Japanese Patent Application Laid-Open No. 56-75139).

The image processing steps of X-ray image data based on such an algorithm are shown in FIG. 9, where each pixel data p of the X-ray image data (the case where the parameter N of the non-sharp mask is set to 9) is shown. Further details will be given with reference to an explanatory diagram and an explanatory diagram showing a microprogram stored in advance in the image processor 7 shown in FIG.

In FIG. 9, i and j are pixel numbers indicating the coordinates of each scanning point.

First, the initial setting start command is sent from the CPU to the CPU.
The data is sent to the image processor 7 via the bus, and the image processor 7 then sets internal calculation parameters (emphasis coefficient β, gradation coefficient γ, mask size N) according to the processing information from the control console (not shown). do.

After completing this set, the image processor 7 continues to hold the microprogram at the same program count position on the microprogram with the next command, a processing start command.

During this time, the microprogram is in a hold state.

Next, the CPU checks the preparation status of the peripheral equipment of this X-ray photography system, and sends a processing start command to the image processor 7 when the preparation is complete. As a result, the image processor 7 reads the X-ray image data stored in the buffer memory 6 and starts arithmetic processing based on equation (1).

This arithmetic processing is continuously executed for each pixel p of the non-sharp mask based on the processing steps on the microprogram, and the processed pixel data is temporarily stored in the buffer memory 6.

In order to continuously perform such arithmetic processing and output of processed pixel data for each pixel, a store (PUSH) command and a return command are provided on the microprogram. 1 in advance as this store command
By setting the number of pixels p per line,
The image processor 7 continuously executes arithmetic processing and outputs the processed pixel data by the set number of times, and when these processes are completed, the microprogram is loaded (POP) to return to the original flow.

When the pixel processing of the l line is completed as described above, the line switching command moves to the pixel processing of the next line, and in the same manner, the last pixel data p, that is, S(1°j) Perform pixel processing up to pixel data.

Note that the time required for line switching is negligible compared to the time required for pixel processing.

In this way, pixel processing is performed in the image processing device 3, and two types of emphasis coefficients β1. β2 and two types of gradation coefficients γ1. The X-ray image data subjected to frequency emphasis processing and gradation processing using γ2 is converted into an optical signal by an X-ray image reproducing device 4, and is further converted into a recording material X by an X-ray image storage device 5. Two processed images AI and B having different emphasis coefficients β and tone coefficients T as shown in FIG. 11 above are formed.

[Problems with background technology]

In the conventional device described above, the coordinates S (i - (N-1) /2.j (N
When processing the pixel data at the point -1)/2) using the parameters of the NXN non-sharp mask, the above processing is performed only after the image processor 7 completes the pixel processing of the pixel data p at the coordinate S (t+j). The unsharp mask signal SUS in equation (1) can be calculated, and based on this calculation result, two types of emphasis coefficients β1. β2. Frequency emphasis processing and gradation processing are performed using two types of gradation coefficients Tl + Tt, and as a result, a sheet of recording material X as shown in Fig. 11 is obtained.
, on which two processed images A, ,B are formed.

However, in this case, only one sheet of recording material X is used. Since the parameters of the non-sharp masks of the processed images Ar and B+ formed above are both N, the contrast of both processed images (&A, , B) is the same, and if these two images A+, B+
In order to make the contrast of the images different, two Batza memories and two image processors are required, which complicates the configuration of the image processing device and reduces processing efficiency and throughput. Ta.

In addition, the image processed image A1. It is also possible to process the B+ parameters using different parameters N+ and Nz for each line, but in this case, the processing time would be nearly twice as long as when the mask size was N, and the processing efficiency would still be poor. There's a problem.

[Purpose of the invention]

The present invention has been made in view of the above circumstances, and it is possible to form two processed images with different contrasts on a single recording material by performing calculations at a speed close to the processing time when the mask size is N. The object of the present invention is to provide an image processing apparatus that can contribute to improving diagnostic performance and improve processing efficiency and throughput.

[Summary of the invention]

The outline of the present invention for achieving the above object is to create an original image signal of visible radiographic image data on a recording material based on radiographic image data obtained by transmitting through a subject in a radiographic system, using So, an emphasis coefficient. is β, and the tone coefficient is T
In an image processing device that can send out processed image data that has undergone frequency emphasis processing and gradation processing on a recording material by performing the calculation expressed by the calculation formula % formula %)], two images can be processed. It has a data processing means and a temporary storage image data memory that is jointly accessed by both of the image data processing means, and when obtaining the unsharp mask signal SUX, two different mask sizes are set and the two image data are processed. Two unsharp mask signals 5IISI*5uSZ are obtained by means of means, and these unsharp mask signals Sus+ + 5us
The present invention is characterized in that two processed image data obtained by subjecting t to frequency emphasis processing and gradation processing, respectively, are sent to the recording material.

[Embodiments of the invention]

Examples of the present invention will be described in detail below.

FIG. 1 is a block diagram showing the configuration of an image processing device 3A of this embodiment, including a buffer memory 6A which is a temporary storage image data memory similar to the conventional device, and X-ray image data stored in this buffer memory 6A. It is equipped with a master image processor 7A and a slave image processor, which are image data processing means that process images according to an image processing algorithm and control a processing microprogram to be described later based on commands from the CPU.

As shown in FIG. 2, the master image processor 7A includes a microprogram memory and a pipeline register (hereinafter simply referred to as "pipeline register") in which a processing microprogram to be described later is stored in advance, and which sends out a microcode (one instruction of the processing microprogram). (referred to as "microprogram memory") 8A and CPU to CPU
A sequencer 9A that accesses the microprogram memory 8 according to commands sent via the U bus and controls the processing microprogram, and a sequencer 9A that controls the processing microprogram by accessing the microprogram memory 8 according to commands sent via the U bus.
10A, which executes calculations on the image data stored in the buffer memory 6A according to microcode instructions sent from the buffer memory 6A. This arithmetic unit 10A is equipped with four arithmetic arithmetic functions of addition, m-arithmetic 9 multiplication, and division so as to be able to execute the arithmetic operations based on the above-mentioned equation (1).

On the other hand, the slave image processor 7B is also substantially similar in configuration to the mask image processor 7A, and includes a sequencer 9B, a microprogram memory 8B, and an arithmetic unit 10B.
However, as will be described later, the microprogram memory 8B stores a microprogram with different contents from the microprogram memory 8A.

Incidentally, a large number of microcode instructions from the microprogram memories 8A and 8B are stored in a buffer memory 6A, respectively.
It is also sent to the microprogram memories 8A, 8
B to sequencer 9A.

Signals for controlling the sequence flow of this device are fed back to each of the devices 9B and 9B.

The microcode instruction group has a width of, for example, 32 to 64 bits, and each bit serves as a control signal for controlling the arithmetic unit 10A and buffer memory 6A via IE. Also, an image processor 7A and a buffer memory 6
In order to operate in synchronization with A, a clock of several hundred nanoseconds is provided, and this clock is controlled by the CPU to allow the processing microprogram to flow.

Also, although it is not shown, is it a mask image processor? A,
The slave image processors 7B each have a large number of enhancement coefficients β5. It has a built-in β table including β2... and a T table including a large number of gradation coefficients 7++rz...

Next, the image processing step of the X-ray image data by the image processing apparatus 3A having the above configuration is performed using each pixel data p (unsharp mask parameter N t ) of the X-ray image data in FIG.
9, when Nz = 3), an explanatory diagram showing the processing microprograms stored in the microprogram memories 8A and 8B, respectively, in FIG. 4, and a processing procedure for X-ray image data in FIG. The explanation will be given with reference to an explanatory diagram showing the image.

In addition, the coordinate S (i − (Nt 1)
/2. J-(Nl-1)/2) and the coordinate S in Figure 8
(i - (N-1) /2. J (N-1)
/2) shall be the same.

The difference between the processing microprogram shown in FIG. 4 and the one shown in FIG. 10 is that the mask image processor 7A and slave image processor 7B access the buffer memory 6A separately within the same basic pixel processing loop. , and a microprogram memory 8A for executing a process for the parameter N1 on the input side of the mask image processor 7 and a process for the parameter N2 on the slave image processor 7B side.

It is stored in 8B.

The buffer memory 6A also includes a register NIT (not shown) for storing accumulated data of the vertical register N, SR, and NI XNI of the parameter N for the mask image processor 7A as shown in FIG. For 7B, there are provided a register N2T (not shown) for storing the accumulated data of the vertical register N15Rt of parameter N and NzxN, and an image memory (not shown). This image processing device 3A basically processes XvA image data using the same image processing algorithm as the conventional device shown in FIG. That is, first, an initialization start command is sent from the CPU to the sequencers 9A and 9B, and each sequencer 9A and 9B stores the instruction address in the microprogram memory 8A.

Send each to 8B.

According to this instruction address, each microprogram memory 8
A and 8B are reoperators 10 for each microcode instruction.
and the buffer memory 6A, whereby the arithmetic unit 10 receives internal calculation parameters (emphasis coefficients β1, β2, and gradation levels for the master image processor 7A and slave image processor 7B, respectively) according to the processing information from the control console. Coefficient rl+12
, unsharp mask parameter Nl =9. Nz = 3
) is set.

In this case, the larger value of the parameters N, , N is set on the master image processor 7A side. Note that the internal calculation parameters are determined by the CPU
It is also possible to set it directly to the recalculation unit 10 from there.

When this internal calculation parameter setting is completed, the sequencer 9'A on the master image processor 7A side continues to wait for a processing start command, which is the next command of the processing microprogram, at the same program count position on the processing microprogram.

On the other hand, the sequencer 9B on the slave image processor 7B side continues to wait for another processor go command at the same program count position on its processing microprogram, while the processing microprogram is in a hold state.

Next, the CPU checks the preparation status of other peripheral devices, and sends a processing start command to the sequencer 9A when the preparation is complete.

As a result, the master image processor 7A sends another processor go command to the slave image processor 7A. Therefore, the processing routine can be entered simultaneously from the next processing step.

The processing microprogram for the master image processor 7A advances the processing steps for the parameter N1 and starts arithmetic processing based on the above equation (1).

This calculation process is first performed continuously in the pixel direction of the first row of the non-sharp mask. In this case, by storing (PUSH) and setting the number of repeats in the sequencer 9 in advance, the processing microprogram continuously executes pixel processing for the set number of repeats, and the processed pixel data is temporarily stored in the Batza memory 6A. Remember.

On the other hand, the processing microprogram for the slave image processor 7B advances the processing step for the parameter N2 and is ready to execute the same pixel processing as in the case described above, but at this time, the sequencer 9B is preset with the parameter N1.
By limiting the coordinates of the unsharp mask for executing the processing step for parameter N2, the pixel processing in the first row on the master image processor 7A side is skipped in the processing step for parameter N2. The program is line switched to pixel processing on the second line.

In FIG. 5, which conceptually shows the processing procedure of the master image processor 7A and the slave image processor 7B, the above-mentioned processing procedure belongs to the I stage, and the basic loop of l pixel processing is repeated a specified number of times, and then the process proceeds to the next stage. Move to.

Here, the coordinates for performing the processing step for parameter N2 are S (i − (NI ri / 2°J
-(N-1)/2) as the center and S(i (NI+
1)/2. J-(N,+1)/2)~S(i-(N
I Nt )/2. J (NI Nz)/2)
The following explanation will be given assuming that it is set to .

Similarly, only the processing steps for the parameter Nl are executed for the second to sixth lines of the non-sharp masks of N and XN, and these results are temporarily stored in the buffer memory 6A.

At the stage of pixel processing on the 7th line, the processing flow shifts to stage 2 shown in FIG. 9 pixel processing) is (
After execution of N, ~NZ) times, the processing step loop for parameter N2 (N2=3 in this example) is executed 3 times.
pixel processing) is master image processor 7A
This is executed by an another processor go command from to the slave image processor 7B.

At this time, the processing step for parameter N2 is executed based on S (i (NI +1) / 2.
J-(N,+1)/2)~S(i (NI N2)
/2. J-(Nl 1 N,)/2) pixels at each point,
It's about data.

Similarly, the parameter N is shown in the 8th and 9th lines.

Nine pixel processes are executed for the parameter N2, and three pixel processes are executed for the parameter N2, and these processed pixel data are temporarily stored in the buffer memory 6.

The processing for parameter N2 is executed only after the processing for parameter Nl starts (Nl-Nz).
After the row is delayed, the state shown in FIG. 3 will be maintained when stage (1) processing is completed.

In this way, 81 pieces of processed pixel data for the parameter N1 and 9 pieces of processed pixel data for the parameter N2 are obtained from each pixel data of the unsharp mask, and these are temporarily stored in the bathophore memory 6A.

Specifically, the accumulated vertical data of parameters N, , Nt is the master image processor? A, the specified number of pixels for each slave image processor 7B, and the data of the non-sharp mask of N + X N 1 and the data of the non-sharp mask of N, It is.

Next, the process moves to the stage (2) shown in FIG.

Based on the instruction address from the sequencer 9A, the microprogram memory 8A sends a microcode to the arithmetic unit 10, which causes the arithmetic unit 10 to perform paraarithmetic processing, βI (So
Sus+) multiplication process, (S.

+β+ (So 5us1)) frequency emphasis processing,
S, = γ, (S0+β+ (So 5usI)
) gradation processing is performed sequentially. When performing this process sequentially, the NI XNI data is calculated as follows.

(a) Register N+ from the accumulated data of register NIT
The data in S R(i NI ) is subtracted and the result is stored in register NIT.

(b) S(i) from the data in register N and SRi.

j−N,) and store the result in register NI.
Store in SRi.

(c) Add the data of S (i, j) to the data of register N and SR4, and store the result in register NI SR4.

(d) Add the data of N and SRi to the data of register NIT, and store the result in register NIT.

Further, in response to another processor go command from the mask image processor 7A to the slave image processor 7B, the slave image processor 7B similarly obtains an unsharp mask signal 5ust corresponding to the coordinates for the processed pixel data for the parameter N2, Next, subtract processing of (365 usz), βt (So 5
multiplication processing of C5o+βz(SoS2)], frequency emphasis processing of St=γ2[S,+βt(So
5 usz)).

By the way, as is clear from FIG. 4, both microprograms execute the basic loop for processing one pixel in a synchronous state, and do not access the buffer memory 6A at the same time.

Further, each processing step is arranged so that the basic loop for processing one pixel in both microprograms can be executed in the same processing time as the microprogram shown in FIG.

In this way, in the image processing device 3A, the emphasis coefficient β
, two types of X-ray image data with different gradation coefficient γ and non-sharp mask parameter N are formed, and as a result, as shown in FIG. Two processed images A and B with different processing and contrast can be formed.

The present invention is not limited to the embodiments described above, and various modifications can be made within the scope of the invention.

For example, the parameter N1 of the unsharp mask. Nz to N+=
By setting Nt, it is also possible to form a processed image similar to that of the conventional apparatus.

〔Effect of the invention〕

According to the present invention described in detail above, two image processors share one buffer memory, so that the same image is processed with different sizes of non-sharp masks on a single sheet of recording material. Two images can be formed within the same processing time as the conventional device. Therefore, it is possible to provide an image processing apparatus that allows two images of the same image having different contrasts to be viewed simultaneously on a single sheet of recording material, and that can contribute to improving diagnostic performance.

In addition, since processing can be performed with different sizes of non-sharp masks at the same time, it is possible to improve the processing efficiency of radiation image data and increase the throughput of the entire system.

[Brief explanation of the drawing]

FIG. 1 is a diagram illustrating an embodiment of the present invention. 2 is an enlarged block diagram showing the details of the apparatus shown in Fig. 1, Fig. 3 is an explanatory diagram showing the coordinates of each pixel of X-ray image data, and Fig. 4 is an enlarged block diagram showing the details of the apparatus shown in Fig. 1. FIG. 5 is an explanatory diagram showing a stored microprogram. FIG. 5 is an explanatory diagram conceptually showing the processing means of the master image processor and slave image processor in the apparatus shown in FIG. 1. FIG. 6 is an explanatory diagram showing the processing means of the master image processor and slave image processor in the apparatus shown in FIG. 7 is a block diagram of an X-ray photography system, FIG. 8 is a block diagram of a conventional image processing device, and FIG. 9 shows each pixel of X-ray image data. Explanatory diagram showing the coordinates of
FIG. 0 is an explanatory diagram showing a microprogram stored in the apparatus shown in FIG. 8, and FIG. 11 is an explanatory diagram showing a processed image on a recording material obtained by the apparatus shown in FIG. 3A... Image processing device, 6A... Buffer memory,
7A... Master image processor, 7B... Slave image processor, 8A, 8B... Micro program memory, 9A, 9B... Sequencer, 10
...multiplier. Agent Patent Attorney Noriyuki Chika Yudo Norio Ogo Figure 3 Figure 7 Figure 8

Claims (1)

[Claims] When forming a visible image on a recording material based on radiographic image data obtained by transmitting through a subject in a radiographic system, a non-sharp mask corresponding to each pixel data of the radiographic image data is provided. The signal S_U_S is obtained, and when the original image signal of the radiation image data is S_0, the emphasis coefficient is β, and the gradation coefficient is γ, the calculation represented by the calculation formula S=γ [S_0+β(S_0−S_U_S)] is performed. In an image processing apparatus capable of transmitting processed image data obtained by subjecting a recording material to frequency emphasis processing and gradation processing, two image data processing means and a temporary storage image that is jointly accessed by both image data processing means are provided. a data memory, and the unsharp mask signal S_
To obtain U_S, two different mask sizes are set, and two unsharp mask signals S_U_S_1 and S_U_S_2 are obtained by both image data processing means, respectively.
And these unsharp mask signals S_U_S_1, S_
An image processing apparatus characterized in that two pieces of processed image data obtained by subjecting U_S_2 to frequency emphasis processing and tone processing, respectively, are sent to a recording material.