JPH02282869A - Image processor - Google Patents

Abstract

PURPOSE:To attain accurate connection processing and the display of images over a wide range with no intervention of an operator by deciding the propriety of plural sheets of images and performing the connecting processing only when the connection is possible. CONSTITUTION:Two images a21 and b22 to be connected together consist of many picture elements, and the areas of the same form and the same size, i.e., the concerned areas are secured in both images. Then both concerned areas are compared with each other for decision of the connecting position between both images (a) and (b). At the same time, the concerned areas of two sheets of picture information are compared with each other and the degree of correlation between both information is outputted. When this output satisfies the prescribed conditions, a connection deciding means 154 produces the detection output. Based on this detection output, a connection means 155 connects both image information to each other at a position where the concerned areas of both information overlap with each other. This connection process is repeated for secure connection among plural images. Thus it is possible to secure accurate connection among plural images with no intervention of an operator.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing apparatus, and more specifically, to an image processing apparatus that performs processing for connecting a plurality of images.

[Prior Art 1] In the case of a conventional imaging device, for example, an imaging device in a real-time display ultrasonic diagnostic device, the size of an obtained tomographic image is limited depending on the device. Therefore, if you want to observe a part that exceeds the size of the obtained tomographic image, you should either line up multiple tomographic images and visually judge the part before inputting it into the imaging device, or manually input the tomographic image to an image processing device. And f-! IIJl operation was used to combine multiple images into one image and display it.

Furthermore, although there is a device described in Japanese Patent Application Laid-Open No. 63-49885 that automatically performs a series of these manual operations, even in this case, operator intervention was required for rough positioning between images.

[Problems to be Solved by the Invention] The combination or connection of a plurality of images by the above-described image processing device is based on the sense of the operator. Therefore, a series of operations for connection are complicated and require skill. In addition, the processing method for accurate positioning is also
A process is performed in which the correlation between the overlapping portions of the two images is calculated, and the two images are superimposed and displayed at the point where the correlation is the highest.

Therefore, in order to make a connection, there must be an overlapping portion between the two images. Conversely, even if there is no overlap between two images, the device will automatically
The correlation between the two images is taken and the two images are forcibly superimposed. Therefore, there was a major problem in that the rough positioning of the operator greatly affected subsequent operations.

In order to solve these problems, it is an object of the present invention to provide an image processing device that automatically and accurately connects a plurality of images with almost no operator intervention.

[Means for Solving the Problems] An image processing apparatus according to the present invention includes a storage means for storing image information of two images, and a region of interest for setting a predetermined region of interest in the image information stored in the storage means. A comparison means that compares regions of interest in two images of the setting means and one image information and outputs the degree of correlation between the two, and a connection judgment that outputs a detection output when the output of the comparison means satisfies a predetermined condition. means, a connecting means for connecting image information of two images based on the detection output in a positional relationship in which the regions of interest overlap, and generating an image as a result of the connection; a display means for displaying the resulting image; and a memory. means, region of interest setting means, comparison means, connection I
The apparatus further includes a control means for controlling the disconnection means, the connection means, and the display means, and outputs the resulting image to the display means.

The image processing apparatus according to the present invention also includes a storage means for storing image information of a plurality of images, a region of interest setting means for setting a predetermined region of interest in the image information stored in the storage means, and A comparison stage that compares regions of interest for two of the images and outputs the degree of correlation between the two images, and a connection "EI + disconnection stage that outputs a detection output when the output of the comparison means satisfies a predetermined condition. Based on the detection output, the image information of the two images is connected in a positional relationship where the regions of interest overlap,
a connecting means for generating an image of the connected result, a display means for displaying the resulting image, a storage means, a region of interest setting+stage,
controlling the comparison means, the connection determination means, the connection means and the display means, and obtaining a resultant image for the image information of the two images;
Control for outputting the resulting image to the display means by performing a connection operation on the resulting image and the image information of another image stored in the storage means, and repeating this for the image information stored in the storage means. means.

[Function 1 According to the present invention, information on a plurality of images is recorded/lf! hair rΩ
A region of interest is set in the image information by a region of interest setting means. A comparison means compares the regions of interest of the two images 111 and outputs the degree of correlation between the two. When the output satisfies a predetermined condition, the connection determining means generates a detection output, and based on this, the connecting means connects the two pieces of image information at a position where their regions of interest overlap. By repeating this connection, multiple images can be connected.

[Example] Embodiments of the present invention will be specifically described with reference to the drawings.

Conventional image processing devices require operator operations to connect multiple images and combine them into a single image.In this embodiment, the conventional image processing device has a connection TII disconnection section and an image connection section. Automate image processing by adding The connection is determined by comparing the difference with a preset threshold.

First, the basic principle of the present invention will be explained with reference to FIG. ? ! To process a number of images as one continuous image, we perform image connection processing on two of those images, and repeat this process for all images. A method for connecting and processing two images will be explained. The relative positional relationship between the two images to be connected includes not only parallel movement in the up, down, left and right directions, but also factors such as rotation.

Here, for the sake of simplicity, the description will be made assuming that the positional relationship between the two images is resolved only by parallel movement in the left and right direction.

In FIG. 1, two connected images a21. and image b 22 is composed of a large number of pixels.

Generally, one pixel has various levels of brightness, but here, for the sake of simplicity, it is assumed that it is represented by a binary brightness signal fO,11. A region of the same shape and size, that is, a region of interest, is provided in each of these two images, and the connecting position of images a and b is determined by comparing the two regions of interest.
3. Also, the image side is set to +10Ib24.

Note that the region ROIa is fixed at a preset position within the image a, but the ROIb can be set at any position within the image.
Since the two images are compared by horizontal translation, the setting position of the region noib is moved by the same height as the region ROIa.

While shifting the set position of the region of interest ROIb by a certain unit pixel within the image, the regions ROIa and l at the shifted position are
The correlation value of l0Ib is determined. The correlation value may be determined by accumulating the absolute difference 11α of the luminance, that is, the tone value, of each corresponding pixel in the regions of interest F101aj; and 1101b.

If this correlation value is calculated while shifting the region of interest R (ltb) within the range that can be set within the image, a correlation value distribution as shown in Figure 2 will be obtained. In the figure, at position: 31
is a point at which there is no relative shift between the two images, and 32 is a point at the peak value of the correlation value distribution. This peak value point 32 becomes a parameter for determining actual image connection.

In order to perform connection processing on two images, the phase 1.1 shift between the two images is determined from the correlation value distribution obtained in this way, and the images are displayed separated by the amount of the shift. You will need to perform operations such as loading the data into memory.

To explain how to connect two images with reference to Figure 3,
In FIG. 1al, images a41 and b42 have a width of 8 pixels, and a region of interest ROEa 43 of width and pixels is set in image a at a position of n pixels from the left end. territory i [RO
Ib is shifted by 1 pixel by 8 from the left end to the right end within the image, and the correlation value is calculated each time. The same figure (bl
shows the distribution of correlation values with respect to the deviation between the regions of interest ROIa and ROIb. The point m where the correlation value is rQJ is the appropriate connection position. FIG. 4(c) shows an image resulting from connecting the image 1 to the left side of the image 1 at this connection position.

In the connected display, the two images will be smaller in the range of js-m+n1 pixels9 For this city part, image a
How to display either J, '5 or [],
A possible method is to display the average value of the gradation values of both images.

To determine the location of the peak of the correlation value, the location where the correlation value is closest to "0" is arithmetically selected.

However, if such a method is adopted, even if two images have no connectivity at all, they may be displayed as connected.

Examples are shown in FIGS. 4 and 5.

Image a51 in FIG. 4 fal has circular data, and image b 52 has linear data, but when the correlation distribution is determined, a distribution as shown in fb) in the same figure is obtained. As a result, a connection display is output, and a connection image 56 as shown in FIG. 5C is displayed. Further, regions of interest ROIa 63 b and 110Jb 64 are set in images a61 and b62 having layered parts as shown in FIG. 5 fat, respectively.
When calculating the correlation distribution of ib) in the same figure, for example, noise 6
When 5 occurs, the peak of the correlation distribution due to this is detected and the two are connected and displayed. Figures 4 and 5
In the case shown in the figure, the reliability of the automated system is improved by making it impossible to determine the relative positional shift between the two images.

Therefore, as shown in Figure 6, from the correlation value distribution,
In addition, in order to detect the peak of the correlation distribution based on the difference in correlation values, three threshold values Tl1l-TH3 for the difference are provided for the purpose of accurately determining the relative position shift. For this,
The distribution data in the figure is sequentially scanned with a shift of width W, and when the difference between the correlation values corresponding to the width W becomes equal to these thresholds, a trigger is applied.More specifically, in this example, first, the correlation value is The difference is the threshold T
111, scanning progresses, distribution value fSA2-5A
A first trigger is activated at a position 71 where I) is larger than the threshold value l.A second trigger is activated at a position 72 where (8 mouths 2-5 BI) is approximately "0" in the vicinity of the tertiary peak.

Subsequently, the difference increases from rQJ to Sfl:2-5CI>
A third trigger is activated at position 73, which is T113. If these three triggers are activated in that order, it is possible to detect a relative positional shift between the two input images only if connection processing is possible from the correlation value distribution. Note that the relative position shift is defined as the midpoint between the difference points of the positions where the trigger 2 is applied.

In FIG. 6, the point i82+Bll/2 is the relative position shift.

A scanning flow in the example of FIG. 6 is shown in FIG. In the flow of the same figure, first manually input the data of images a and b, set CIfJOl, and set 1101+i in image a.
1021 Then, it is checked whether the entire set area of the image has been set as the region of interest 11012 (+1041), and if it has been set, the process ends. If not set, move the setting value of region @ROIb one pixel at a time from the setting end +
1061, take the correlation between the region of interest 1101a and ROIIII (1081, determine the difference value (I lo
t. It is checked whether or not the trigger 1 is applied for the difference value 11121, and if it is not applied, the process returns to 104, and if it is applied, it is checked whether the trigger 2 is applied or not fll, 4+
If 1-Riga 2 is not applied, the process returns to step 1υ4, and the applied Kawane determines the connection Ij [6]. Next, check whether trigger: 3 is applied or 100.fl181゜If trigger 3 is not applied, return to step +04,
If it takes, connect and display images a and b (1, 2
0+ End processing.

8th [4] A block diagram of an embodiment in which the present invention is introduced into an image processing device is shown. In the figure, an image processing device 80 according to the embodiment has a connection↑II disconnection section 154 and an image connection section 156 connected to the conventional device, and each section 151 to 156 is connected via a data bus 150. system (1 part 15
1 is a control unit that controls the human output of the image data of the valley; 6; the image input unit 152 is a human-powered device that reads image data composed of brightness signals of convex pixels for each of a plurality of images; The image storage unit 153 stores the read image data.

The 't'll THj unit 154 is a functional unit that determines whether or not two images can be connected, and when it is possible to connect the two images, outputs a connectable signal indicating that. The image connection unit 155 is a processing unit that performs a connection process to connect two image data when a connection enable signal is output, and the image display unit 15G
is an image display device that displays a connected image using the resulting image data.

A specific example of the configuration of the connection t'll disconnection section 154 is shown in FIG.
2 [) has 4. tl (] I a setting section 20 (
) is a storage unit that inputs and stores the image data of the region of interest RO[a set in the image a connected from the image f■ input unit 152. The RO[b setting unit 201 stores image data of a region norb that is manually input one after another each time the region of interest ROIb of the other image to be connected moves.

The correlation value calculation unit 202 calculates the region of interest 11 from the setting unit 200.
The image data of 0Ia, and the setting section R01b from February
This is a calculation unit that manually processes the image data of , calculates and outputs the correlation value between the two fronts. The difference value calculation unit 2D:l is a calculation unit that manually inputs this output, calculates and outputs the difference between the correlation values. The trigger detection unit 204 has a detection function that inputs the difference and compares it with a threshold value to detect the presence or absence of a trigger.

If the trigger is not activated, the ROIb setting section 201
The same process is performed by shifting the position of the region of interest tpinorb.

When a trigger is applied, a trigger signal is output, and this signal is sent to the connection determining unit 154 via the control unit 151.6 Also, the image on which the connection process has been performed is stored in the image storage unit +53, and the image and By transferring a new image to the ROI setting units 200 and 201 and performing the above-described processing, it is possible to connect a larger image. By repeating this operation, it is possible to connect multiple images to create one large image.

The explanation so far has been about the case where the gradation value data of the image takes two values of "0" and rlJ, but even when the gradation value data takes level 3 or higher, the gradation value between pixels Similar connection processing is possible by summing up the absolute values of the differences and using the correlation value in the determination means for applying the trigger.

Further, when there are 77 pieces of data that differ from the background at only one point in one image, such as images a300 and b301 shown in FIG. 1O (circular portion in the figure).

Two images in which the region of interest ROIa is set at position 302 in image a can be connected. However, if this is set at the 1st position circle 3, it becomes impossible to connect the images because it is difficult to distinguish it from other backs 1 in the image. In order to deal with such a case, the position setting of the region of interest 1t01a is set to an operator's manual operation mode, so that the setting can be changed from position 103 to a position such as 31)2.

The case where the setting value of the region ROIa is fixed in advance is called the "fixed setting method", and the case where it is set manually is called the "arbitrary setting method"6. In the fixed setting method, the region of interest ROIa is set at the center of the input image.
is set because it is assumed that there is no information as to whether the image is shifted to the left or right with respect to the image a. If this information is known, in order to detect a larger relative position shift, if the image is shifted to the right with respect to image a, set the setting position of region ROIa to the right of image a, If it is shifted to the left, area 10
The setting position of Ta may be set to the left side of the image screen. The case where the set value is corrected when it deviates to the right side is the "right self-limited, fixed setting method", and the left 1i1. The correction when the deviation is +1 will be referred to as the "leftward limited setting method."

In this embodiment, the operator can select one of these four methods for positioning the group of interest tjiROIa. The ROIa setting section 90 has a configuration as shown in FIG.

The operator manually inputs an operation selection command to the image processing device 80, and this command is sent to the image processing unit 152 or the image storage unit 153.
The information is sent to the setting control section 320 from there. The turning section 320 controls the fixed setting section 321 when the command is a fixed setting, the arbitrary setting section 322 when the command is an arbitrary selection, the left direction limited setting section 323 when the command is for setting only to the left, and the right direction. At the time of setting, the right direction limit setting section 324 is connected to execute the command.6 The conditions of interest ti4 are that ROIa and ROIb have substantially the same shape and number of pixels. But they are
It does not have to be rectangular like the examples shown so far. Further, when the region of interest ROI is configured from a plurality of parts, there is an advantage that a wide image area can be set as the region of interest ROI with a small number of pixels.

If the positional relationship of the image (I(1) cannot be resolved only by parallel movement in the horizontal direction, but also includes a translation component in the vertical direction, the series of operations described in Fig. 3 is performed to set the region of interest ROIb. It can be processed by changing the height until the same ROfb is set for the entire area in the image.Also, if rotational elements are included, the region of interest norb is set every - times in the horizontal direction. It can be processed by rotating it little by little, moving it in parallel, and detecting the peak value.

FIG. 12 shows an image processing device 82 according to another embodiment of the present invention. This device 82 is an image data memory 402
and an image display memory 405, and the CPU 401 uses each routine of the program stored in the program memory 409 to perform image information recording/setting means, region of interest setting means, correlation degree comparison means, and image displaying means. It can be operated as a connecting means or a display means.

In the image information storage routine, a plurality of image data are input to the image data memory 402 via the data bus 408 from an image input section (not shown). In the region of interest setting routine, the address corresponding to the region of interest in the image data memory 402 is stored in the program memory 40.
9 is stored. In the comparison routine, the data included in the regions of interest set in each of the two images stored in the image data memory 402 are compared using a method similar to the correlation degree comparison method described above.

In the image connection routine, 1 image data memo/40
The image in 2 is the data bus 408 and gate circuit 403
The image is transferred to the image display memory 405 via . Here, the two images are stored in a positional relationship such that the connection point determined by the comparison routine takes the same address in the image display memory 405.In the 6 or 5 image display memory 405, there are two images. Data of one image in which images are connected is stored.

In the display routine, the image data in the image display memory 405 is displayed on the scanning line (
raster) and are read out sequentially in synchronization with the gate circuit 404.
and the DA converter 406, each pixel data is brightness-modulated and displayed as a connected image on the display 23407.

In the 0 image data memory 402, the details of the operation of the image connection routine will be described below, data of an image a of mxn pixels is stored at storage locations fXa1. In the area with the diagonal from Yll to (Xam, Yn), the data of the same < mxn pixel image is stored at the storage location (Xbl, Yll-fXbm, Yn).
Assume that the data are stored in areas diagonally opposite each other.

Considering the case of movement only in the X direction as in the above example, the connection point detected by the comparison routine stored in the program memory 409 is Xaj in image a.
, image A is assumed to be xbJ, and image A is connected to the right side of image a.

Now, in the write cycle of the 5-image display memory 405,
cpυ401 is data D1Xa1. of image a. Yl)～
D (Xaj, Ynl connected to data bus 408, gate circuit 403, FIG. 14 fb) image display memory 405
Address (old, Vll ~ (Uk-1, Vnl), then transfer the image data DfXbf, Yll-
D (Xbm, Ynl are similarly stored in the image display memory 40.
5 (7) 7 L/, 2. fUk, Vl) ~
fU1. Vn) Transfer. As a result, Fig. 14 fb
As shown in FIG. 1, image data in which image a and image S are connected is reconstructed in the image display memory 405. Data Q transferred to these image display memory 405
teeth.

In synchronization with the raster on the display 407, the CPU 401 outputs the first raster: Q(Ul, Vll, QfU2.Vll, QfU3.
Vll, , , , Q (Ue, V11 second raster taco old U1.V2), QfU2. V21. QfU3. V2
),,,,,Q(Lle,V21 third raster Q(Ul,V31.QfU2.V31.Q(U3.
V31. , , QfLIe, V3) are read out in the order up to the n-th raster, inputted from the gate circuit 404 to the DA converter 406, and after digital-to-analog conversion, it becomes a connected image and is displayed on the display 407. Is displayed. However, fUx, Vy) = (x, yl+
xl ~e, y=l ~nfUk-1, Vnl
= fXaj-Xal, Ynl (υ, Vl) =
fXaj−Xal+1. I) (Ko<Vn) −= (
It is assumed that Uk+Xbm-XbJ, Ynl, and e≧14. Further, Ue is the final address in the X direction of the image display memory.

In the above example, an image display memory 405 is provided separately from the image data memory 402, but the configuration may be such that these memories are used in common. In that case, an image processing device 84 according to an embodiment is shown in FIGS. 1 and 3, and CI] II 501 is similar to the embodiment shown in FIG.

Each routine stored in the program memory 510 allows the present device to operate as various means such as image information recording and recording means.

The image memo 'J505 has the above-mentioned image data memory 402.
Assume that image a and image image are written and displayed in the same way. Now, the connection points Xa,j (image a) and Xbj (image b) detected by the comparison routine stored in the program memory 510 are stored in the connection address register 502.
Saved in During the display cycle of the image memory 505, the gate circuit 503 is turned off, the gate circuit 504 is turned on, and the image data is sent to the display 5 via the DA converter 506.
The contents of the 0 image memory 505 sent to the display 50
7 rasters are read out sequentially in synchronization with each other. At this time,
The selection of the X address is based on the contents of the connection address register 502 [j19. Xal, Xa2. Xa3. ,
,, ,Xa,j-2,Xaj-1,Xbj.

Xbj+1. Xbj+2. ,,,, do it in Xbn and Iu order.

As a result, the output data is the first raster: DfXal, Yl), DfXa2. Yll, DfXa
3. YI) DfXa, j-1, Yll, 0(Xbj,
Yll, DfXbj+1. Yl)DfXbj+2. Yl
l, , , , D(Xbn, Yll second raster: DfXal, Y21.D(Xa2.Y2), D(Xa3
．． Y2),,.

DfXaJ-1, Y21, DfXbj, Y21, D
fXb,j+1. Y2)D(Xbj+2.Y21.,,
, , DfXbn, and Y21, and after DA conversion, a single image connected as shown in FIG. 14 (bl) is displayed on the display 507.

FIG. 15 shows how the contents of the image memory 505 in the embodiment shown in FIG. 13 change as a result of this repeated operation. 1jii to be processed is image memo Q505
As shown in FIG. 5A, data for a plurality of images, four in this example, are stored independently from each other.

First of all, a connection is determined for the data of image a and image
It is assumed that when Ib is set, a connectable signal is output. Based on this connection position data, the image memory 50
The content of 5 is rearranged, and the result is the content shown in bl.

Next, the data of image ab and image C, which are the result of connecting images a and b, are determined to be connected in the same way, and if connection is possible, the contents of the 1-image memory 505 are changed to the state shown in the same fct. Placed. This constitutes image abc.

Similarly, when images abc and image d can be connected, the image memory 505 is rearranged as shown in fdl,
This completes the repeat operation. An image represented by the image data completed in the image memory 505 is displayed on the display 507.

An example of a processing flow for connecting three or more images is shown in FIG. 16. The processing flow in the same figure differs from that shown in FIG. It differs from In step [25, the data obtained by connecting the two images ai3 and b are stored in the image memory 505.
, and returns to step 123 in a loop of interlace symbol 13. In step 123, it is determined whether there is image data to be connected, and if there is no image data, step 123 is performed.
Execute and exit, but if there is one, repeat the same connection process for it. The other processing may be almost the same as the processing for the same reference 11α code in the flow of FIG.

[Effects of the Invention 1] As detailed above, 5 the present invention determines whether or not multiple images can be connected in an image processing device, and performs connection processing only when connection is possible, so operator intervention is not required. This has the effect that images of a wide area can be connected and displayed accurately without any need for connection.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram illustrating the positions of regions of interest set in two images in an embodiment of the present invention, FIG. 2 is a diagram showing a correlation distribution between regions of interest, and FIG. 3 (at, (bl) and fcl are explanatory diagrams showing two images that can be connected, their correlation distribution, and connected images, respectively; FIG. 4(a); fbl and fcl are two images that cannot be connected, their correlation distribution, and connected images; Figure 5 fa) and Figure 5 (b) are diagrams showing two images with noise and their correlation distribution, respectively. Figure 6 is the threshold Rti and trigger for detecting the peak of the correlation distribution. FIG. 7 is a flowchart showing an example of the operation flow of peak detection in the example of FIG. 6, FIG. 8 is a functional block diagram showing an embodiment of the image processing device according to the present invention, FIG. 10 is a functional block diagram of the connection determination unit in the embodiment; FIG. 1O is a diagram showing connectable and unconnectable positions of the region of interest; FIG. 12 and 13 are functional block diagrams showing other embodiments of the present invention, and FIG. 14(a)8 and fbl show internal layouts of the image data memory and image display memory in these embodiments, respectively. An explanatory diagram showing an example, Fig. 15 is 5, 1
FIG. 4 is an explanatory diagram showing how the contents of the image memory change in the embodiment shown in FIG. 3; FIG. 16 is a flow diagram similar to FIG. 7, showing an example of a processing flow for connecting three or more images. Explanation of ':' in main part 80.82.84. Image processing device 90, region of interest setting units 151 and 152. 153, , , . 154, , , . 156, , , . 401.501. . 402, , , . 405, , , . 406.506. . 407.507 409.510. . 505, , , .

Claims (1)

[Scope of Claims] Storage means for storing image information of one or two images; region of interest setting means for setting a predetermined region of interest in the image information stored in the storage means; Compare the regions of interest for the two images,
a comparison means for outputting the degree of correlation between the two; a connection determination means for outputting a detection output when the output of the comparison means satisfies a predetermined condition; and image information of the two images based on the detection output. connecting means for connecting the two in a positional relationship in which the regions of interest overlap and generating an image of the result of the connection; a display means for displaying the resulting image; the storage means, the region of interest setting means, the comparing means, and the connection determining means. , a control means for controlling a connecting means and a display means, and causing the resulting image to be outputted to the display means. 2. Storage means for storing image information of a plurality of images; Region of interest setting means for setting a predetermined region of interest in the image information stored in the storage means; Regarding two images among the image information. a comparison means for comparing the regions of interest and outputting the degree of correlation between the two; a connection determination means for outputting a detection output when the output of the comparison means satisfies a predetermined condition; a connecting means for connecting image information of two images in a positional relationship in which regions of interest overlap and generating an image as a result of the connection; a display means for displaying the resulting image; the storage means, the region of interest setting means, and a comparison. control means, connection determining means, connecting means, and display means to obtain the resultant image for the image information of the two images, and image information of the resultant image and other images stored in the storage means. an image processing apparatus comprising: a control means for causing the display means to output a resultant image obtained by performing a connection operation for the image information stored in the storage means and repeating this for the image information stored in the storage means. 3. In the apparatus according to claim 1 or 2, the region of interest setting means moves the position of the region of interest set for the image information of at least one image, and the comparing means moves the region of interest at each of the positions. An image processing device characterized by outputting a degree of correlation. 4. The apparatus according to claim 1 or 2, wherein the comparison means calculates the absolute difference in tone values between pixels in corresponding positional relationships within the region of interest set in the image information of the two images. An image processing device characterized by calculating values and outputting the degree of correlation based on the cumulative value. 5. The image processing apparatus according to claim 1 or 2, wherein the connection determining means outputs the detection output by comparing the degree of correlation with a predetermined level. 6. The apparatus according to claim 3, wherein the connection determining means outputs the detection output by comparing an amount of change in the degree of correlation with a predetermined level with respect to movement of the region of interest. Image processing device. 7. An ultrasonic diagnostic apparatus comprising the image processing apparatus according to any one of claims 1 to 6.