JP2995342B2 - Area detection method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention relates to an image processing apparatus, and particularly to an image processing apparatus for designating an area to be edited / processed with a frame line. The present invention relates to an in-frame area detection method for detecting an area (in-frame area) surrounded by a frame line.

(Prior Art) As a method of designating an area for processing and editing an image, a method using a frame line is known (for example, see Document 1: Nishimura et al., Image Magazine vol. 17, PP328-336, 1988). . In this method, a frame representing the boundary of a region is created, and the inside of the frame is designated as a region. Compared with the conventional area specification method using tablet or coordinate input, this method has the advantage that the area of any shape can be easily specified and the area can be specified on the same plane as the original image, making it easy to understand the correspondence with the original image. There is.

In such an area designation method, a technique for detecting an area inside a frame line (an area within a frame) is indispensable. For example, the following method is known as a method for detecting the area in the frame.
First, the outermost portion of the image is set as an outer frame portion, and the presence or absence of a pixel adjacent to the outer frame portion is sequentially checked. A pixel adjacent to the out-of-frame portion is determined as an out-of-frame pixel, a pixel not determined as an out-of-frame pixel is determined as an in-frame pixel, and a region of the in-frame pixel is detected as an in-frame region. According to this method, an area within a frame can be detected by a simple process from an image signal obtained by raster-scanning an image. However, the frame lines that can be detected are limited to fairly simple figures in terms of topology, and the area within the frame is changed from a frame line having a complicated shape such as a so-called concave (a shape having a concave portion on the lower side on the screen) or a multiple frame. It was difficult to detect correctly.

Therefore, a method of detecting an in-frame area using a nest level is also known. This applies a level attribute to the frame portion of the image signal and other portions to generate a binary frame signal, and sequentially indicates the nesting level of each pixel of the current raster from the connection with the previous raster. In this method, a level signal is calculated, and a determination is made as to whether the pixel is inside or outside the frame from the level signal. According to this method, even in the case of a frame line having a shape such as a concave bottom or a multi-frame, a region within the frame can be correctly detected by a simple raster-based process, and the detection stability for the whiskers of the frame line is also high. . However, in this method, it is necessary to simultaneously perform the calculation of the level signal of the current raster and the conversion to the level signal of a connected area (run) unit in which the value of the frame signal is equal and continuous, and the processing is complicated. Since in-frame detection is usually applied to an image signal obtained by raster scanning, such complexity of processing directly leads to an increase in circuit size.

(Problems to be Solved by the Invention) As described above, in the conventional method of detecting a frame region using a nest level, regions within a frame are compared even from a frame line having a complicated shape such as a concave bottom or a multi-frame. On the other hand,
A complicated process of simultaneously performing the calculation of the level signal and the conversion of the level signal into the unit of the connection area is required, and there is a problem that the circuit scale is increased.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method for detecting an in-frame area that can easily detect an in-frame area from a frame line having a complicated shape such as a concave bottom or a multi-frame, and can simplify the processing.

[Configuration of the Invention] (Means for Solving the Problems) In the present invention, a frame signal is obtained by assigning a level attribute to both a frame line portion constituting a frame and a ground portion which is a portion other than the frame line portion. The level signal indicating the nest level is sequentially calculated according to the connection relationship of the above, and the depth of the nest of the ground or the frame is detected according to the calculated level signal, thereby detecting the area within the frame. Further, the calculation of the level signal is divided into the calculation of the provisional level signal referring only to the adjacent pixels of the previous raster and the calculation of the level signal (representative level signal) in the same raster.

That is, the present invention relates to an in-frame area detection method for detecting an in-frame area surrounded by a frame on an image from an input image signal obtained by, for example, scanning (raster scanning) an image including the frame. It is determined whether or not the image signal is a frame line portion for each pixel, and a binary frame signal is generated. With respect to the current pixel (pixel of the current raster) of the input image signal, a previous pixel (previous raster) adjacent to the current pixel is generated. Pixel), a tentative level signal is calculated from the level signal of the previous pixel and the frame signal of the current pixel, and for the current pixel column in which the tentative level signal has been calculated, the connection of the frame signals having the same and continuous values is performed. A region is detected, a level signal (representative level signal) of a pixel in the connected region is calculated from a value of a temporary level signal of each pixel in the connected region, and a frame is calculated from a value of the level signal of a pixel in the connected region. Judging the inside and outside of the line and detecting the area inside the frame And butterflies.

Further, the level signal of the pixel in the connected area is corrected using the relationship between the level signal of the pixel in the adjacent connected area and the frame signal, and the inside and outside of the frame are determined from the corrected level signal value. The inner region may be detected.

Also, since the calculation of the representative level signal of each pixel is equivalent to calculating the lower distance from the image scanning start end for each pixel, the inside and outside of the frame line of the pixel are determined from the value of the lower distance. Is also good. Here, the “downward distance from the image scanning start end” means that the number of movements from the pixels of the frame portion to the pixels that are not the frame line portion in the path that sequentially traces the adjacent pixels from the image scanning start end downward is the smallest. It is defined by the number of movements on the route. "Following down" means moving from a frame pixel to a frame pixel, or from a ground pixel to a ground pixel, in any direction other than the reverse direction of the sub-scanning direction, and moving from the frame pixel to the ground pixel or from the ground pixel to the frame pixel. When moving to, this means tracing in the sub-scanning direction.

(Operation) By calculating the level signals of the frame and the ground in this manner, the nest level of the frame, that is, the number of the frame with respect to the outermost frame, is detected, whereby the double frame or the lower frame is detected. For a frame such as a concave frame, a reasonable in-frame area detection is performed, and an influence such as a "whisker" generated when a handwritten frame is formed is also removed. In addition, when calculating the level signal, the calculation of the provisional level signal referring to the previous raster and the calculation of the level signal referring to the same raster are performed separately, thereby simplifying the calculation process and increasing the speed. Further, when the processing is realized by hardware, the circuit scale is reduced.

(Embodiment) FIG. 1 is a block diagram showing a schematic configuration of an in-frame area detecting apparatus according to a first embodiment of the present invention. The in-frame area detection device includes an image input unit 1, a binarization circuit 2, a temporary level signal calculation circuit 3, a representative level signal calculation circuit 4, and an in-frame / out-of-frame determination circuit 5.

The image input unit 1 reads an image on a document by raster scanning with a line image sensor and converts the obtained image signal into an A / D signal.
After the conversion, the device outputs the data through correction processing such as shading correction. The frame signal generating circuit 2 includes an image input unit 1
Is divided into a frame portion and a ground portion other than the frame signal based on an appropriate standard such as density or chromaticity, and a value of “1” is taken in the frame portion and “0” is taken in the ground portion. A frame signal P (x, y) composed of a value signal is output. However, the frame portion is the portion where the frame is drawn, and the ground portion is the other portion. X and y represent coordinates. The provisional level signal calculation circuit 3, the representative level signal calculation circuit 4, and the in-frame / out-of-frame determination circuit 5 are surrounded by a frame signal from a frame signal by a following algorithm, that is, a frame line written in a specific color on a document, for example. Detected area.

FIG. 2 and FIG. 3 show the algorithm of the in-frame area detection processing in this embodiment. For convenience of explanation, the fourth
As shown in the figure, the origin is the upper left corner of the image, the x axis is right (positive on the right), the y axis is positive (down on the right), and the range of the image is 1 ≦ x ≦ x max, 1 ≦ Let y ≦ y max. Also, x
One line in the axial direction is called a raster, and the raster whose y coordinate is y0
12 is called a y0th raster. Also, the y coordinate is 0 at the top of the image.
Of the image is referred to as the upper end of the image. Hereinafter, all the drawings used in the description use this coordinate axis.

As shown in FIG. 2, in the present algorithm, a level signal L (x, y) representing a nest level is defined for each pixel of the image, and the frame signal and the level signal of the already calculated area are recursively used. Level signals L (x,
Calculate y). That is, first, the level signal L (x, 0) of the raster 11 is set to 0 (step S11), and the frame signal P (x, 0) is set.
1) The level signal L (x, 1) is calculated from P (x, 0) and the level signal L (x, 0) (steps S12 to S13). However, the frame signal P (x, 0) at the upper end of the image is all 0, that is, a ground pixel, and the level signal L (x, 0) at the upper end is also 0. Next, based on the calculated level signal L (x, 1) and the frame signal P (x, 1), it is determined whether the pixel Q is inside or outside the frame (step S14). Next, the calculated level signal L
(X, 1) and frame signals P (x, 1) and P (x, 2),
Calculate the level signal L (x, 2). Hereinafter, similarly, the level signal L (x, y) is calculated with reference to the level signal L (x, y-1) and the frame signals P (x, y-1), P (x, y), and the Y = y max while incrementing y by 1
(Steps S13 to S16).

Calculation of the level signal in the raster, that is, the frame signal P (x, y-1) and the level signal L (x, y-
1) and the level signal L from the frame signal P (x, y) of the y-th raster
FIG. 3 shows details of the algorithm (step S13) for calculating (x, y). This calculation is roughly divided into two main steps. First, as the first main step,
The frame signal of the pixel Q at the coordinates (x, y) is compared with the frame signal of the pixel Q0 at the coordinates (x, y-1) adjacent to the pixel Q, and the combination (see FIG. 5) A provisional level signal L representing a provisional nesting level of coordinates (x, y) according to <Rule 1>
Calculate (x, y). This is performed for each pixel in the x-axis direction (steps S21 to S24).

<Rule 1> (a) When the pixel Q is a frame and the adjacent pixel Q0 is also a frame, the nest level of the pixel Q0 is set as the nest level of the pixel Q.

if P (x, y) = 1, P0 (x, y-1) = 1 then L (x, y) = L (x, y-1) (b) Pixel Q is a frame and adjacent pixel Q0 Is the ground, the nest level of the pixel Q0 is set as the nest level of the pixel Q.

if P (x, y) = 1, P0 (x, y-1) = 0 then L (x, y) = L (x, y-1) (c) Pixel Q is ground and adjacent pixel Q0 Is a frame, the nest level of the pixel Q0 + 1 is set as the nest level of the pixel Q.

if P (x, y) = 0, P0 (x, y-1) = 1 then L (x, y) = L (x, y-1) +1 (d) Pixel Q is ground and adjacent pixel When Q0 is also the ground, the nest level of the pixel Q0 is set as the nest level of the pixel Q.

if P (x, y) = 0, P0 (x, y-1) = 0 then L (x, y) = L (x, y-1) By the above processing, the provisional level of each pixel of the y-th raster When the signal L (x, y) is determined, as a second main step, the relationship between the frame signal of the pixel adjacent in the horizontal direction and the provisional level signal in the raster y according to the following <Rule 2> (FIG. 6) (See step S2).
5 to S26).

<Rule 2> A set of adjacent pixels (x,
y), (x + 1, y), processing 1 for those pixels
I do. This is repeated until there is no pixel set satisfying the condition 1.

Condition 1: The pixel Q2 horizontally adjacent to the pixel Q1 is a frame pixel or a ground pixel, and their levels are different.

Process 2: The level signal of the pixel having the larger level signal of the pixels Q1 and Q2 is changed to the smaller level signal.

In this processing, one raster is divided into connected areas (runs) having the same frame signal value, and the minimum value of the level signal of each pixel in each connected area is represented by a representative level signal representing the nest level of each pixel in the connected area. Is equivalent to

Finally, from the calculated value of the representative level signal L (x, y), it is determined whether the pixel is inside or outside the frame by the following <Rule 3> (Step S14 in FIG. 3).

<Rule 3> If the pixel Q is a ground pixel and its level is an even number, the pixel Q is an area outside the frame. If the level of the pixel Q is an odd number, the pixel Q is an area inside the frame. If the pixel Q is a frame pixel, In this case, the pixel Q may be determined as an out-of-frame area depending on whether or not the frame line is regarded as being in the frame. Here, it is assumed that the portion on the frame is regarded as inside the frame.

<Description of In-Frame Detection by This Algorithm and Processing Example> (Definition of Distance and Downward Distance) First, the “distance” between the ground pixel P and the ground pixel Q is defined as follows. Pixel Q by sequentially tracing adjacent pixels from ground pixel P
Of the routes that lead to, the route with the least number of crossings over the frame is called the minimum oversize route, and the number of times over the frame in this case is two points
Defined as the distance between P and Q. Here, "number of times exceeding the frame"
Means the number of transitions from the frame pixel to the ground pixel. For example,
In FIG. 18, the minimum crossing path from the pixel P1 to the pixel Q1 is L1, and since this path does not exceed the frame portion, the distance between the pixels P0 and Q0 is 0. In addition, the minimum frame passing path from the pixel P2 to the pixel Q2 is L2, and the distance is 1 because L2 crosses the frame once. Similarly, the minimum crossing path and the distance from the pixel P3 to the pixel Q3 are L3 and L2.

As the definition of the inside and outside of the frame figure, here, pixels having an even number of pixels when viewed from an outermost reference point of the image are defined as an area outside the frame,
Pixels with an odd number of distances are considered as an area within the frame. according to this,
The area inside the simple closed curve is the area inside the frame, and the area multiplexed by a plurality of frames is the area inside the frame and the area outside the frame alternately from the outside frame. This definition is considered valid as a definition inside and outside the box.

However, calculating the above definition for a general frame image requires a huge amount of calculation and is difficult to apply to a raster scanning signal. Therefore, in this processing, as a concept based on the above-described minimum frame passing route and distance, the lower minimum frame passing route and lower distance are defined as follows.
That is, of the paths that follow “downward” from the pixel P to the pixel Q below the pixel P (having a large y coordinate), the path with the least number of crossing the frame line is defined as the lower minimum frame crossing path. The number of times exceeding the line is defined as the downward distance. Here, the path that follows “downward” is a path that follows definition 1 below.

Definition 1: A downward path is the same type of pixels (ground pixels or frame pixels) adjacent to the horizontal direction or downward,
In the case of different types of pixels (ground pixel and frame pixel), the path is taken to an adjacent downward pixel.

Then, pixels whose even distance from the upper end of the image is even are defined outside the frame, and odd pixels are defined as inside the frame. The downward distance according to this definition has a concept similar to the distance. In particular, if the frame line is a closed curve, inside the frame by the downward distance, outside the frame is within the frame by the above distance,
It becomes the same as the judgment outside the frame. FIG. 19 shows an example of the lower minimum frame passing path and the lower distance for the actual frame figure. In the figure, the lower minimum frame passing path from the pixel P1 to the pixel Q1 is L1, and the lower distance is “0”. Similarly, L2 and “1” are obtained from the pixel P2 to the pixel Q2, respectively.

(Explanation of Equivalence of Level and Downward Distance) The level signal of each pixel calculated in this process is equivalent to the downward distance from the upper end of the image. The reason will be described. First, consider a pixel P whose distance from the upper end of the image is n below. FIG. 7 shows an example where n = 1. In the figure, each cell corresponds to one pixel, a portion surrounded by oblique lines represents a frame pixel, and the other portions represent ground pixels. By definition, there is a path that reaches the pixel 21 (P) from the pixel 22 at the upper end of the image over the frame line n times (indicated by an arrow column in the figure). Consider the level of each pixel on the path. By assumption, the level of the start point 22 of the route is “0”. In the case where the same type of pixel is traced in the left-right direction on this route 23, the level is equal to or smaller than the level of the previous pixel according to <Rule 2>. In the case where the same type of pixel is traced downward, the level becomes the same as that of the preceding pixel according to (a) and (d) of <Rule 1>. Also, when the route is from the ground pixel to the frame pixel 25, the level is always lower than the definition 1, so that the level does not change from the previous pixel according to (b) of <Rule 1>. When the route moves from the frame pixel to the ground pixel 26, the pixel level becomes 1 higher than the previous pixel by (c) of <Rule 1>. Therefore,
The level of each pixel on the path increases by 1 only when moving from the frame pixel to the ground pixel, and otherwise does not change or decreases, so the level after n times over the frame line portion becomes n at most. . Therefore, the ground pixel P whose distance below the upper end is n
Is n or less. In the figure, the numbers in the pixels indicate the calculated levels.

Conversely, if the level of the ground pixel P is n, the lower distance from the upper end to the ground pixel P is n or less, as shown in FIG. Therefore, it indicates that there is a path from the upper end to the ground pixel Q that reaches the frame pixel n times more than n times. First, consider a connected area 32 of ground pixels including a certain ground pixel 31 (Q) and connected in the same raster. In this area 32, the previous step S2
5. There is a pixel R whose level is n before the correction processing by S26. The pixel Q2 adjacent above the pixel Q is defined as an upper pixel of the pixel Q. When the pixel Q2 is a ground pixel, <Rule 1>
(D), the level of the pixel Q2 becomes n. Also, the pixel
When Q2 is a frame pixel, the level of the pixel Q2 is n-1 according to (c) of <Rule 1>. However, when the level n is 0, the pixel Q2 is always a ground pixel at the level 0. Similarly, consider a connected region of frame pixels including the frame pixel S and connected in the same raster. As before, the steps in this connection area
There is a pixel T whose level is n before the correction processing in S25 and S26. The pixel S2 adjacent above the pixel T is defined as an upper pixel of the pixel S. According to (a) and (b) of <Rule 1>, the level of S2 is n regardless of whether S2 is a frame pixel or a ground pixel.

The upper pixel of the pixel P is P2, and the upper pixel of P2 is P3.
To reach the upper end of the image. The level of the path following these pixels decreases by one each time the path moves from the ground to the frame. Since the level at the upper end is “0”, this path crosses the frame portion n times. Therefore, the lower distance from the upper end to the pixel P is n.

As described above, the fact that the level of the ground pixel P is n is equivalent to the fact that the downward distance from the upper end of the image to the ground pixel P is n. That is, the level determined by this processing is
Represents the downward distance from the top of the image.

Next, with reference to FIGS. 9 to 14, specific examples of nest level calculation and in-frame / out-of-frame determination for the frame lines of some shapes will be described with reference to FIGS. Indicates that there is consistency with the area within the frame.

(Circle Frame Line) First, an example of a simple circular frame (FIG. 9) will be described. In the figure, the solid line or hatched portion represents the frame portion,
The number surrounded by ○ represents the level of the ground pixel, and the number surrounded by □ represents the level of the frame pixel. The following examples are similarly represented. Since all the pixels outside the frame line 41 have a path that reaches from the upper end of the image without passing through the frame, the level is “0”, and the area is outside the frame. Further, since the inside of the frame line needs to cross the frame line at least once from the upper end of the image, the level is "1", and it is determined that the area is within the frame.

Next, an example of a frame (FIG. 10) having a concave portion on the lower side will be described. Of the pixels outside the frame 51, the lath 52 and the frame 51
The level other than the area sandwiched between the bits is “0”. To reach the area inside the frame 51, it is necessary to cross the frame at least once from the upper end, so that the level of this area is "1". To reach the area between the frame 51 and the raster 52,
Since it is necessary to cross the frame twice, the level of this area is "2". Therefore, only the inside of the frame line whose level is an odd number is determined to be a region within the frame.

(In the case of multiple frames) When a plurality of frame lines are multiplexed, it is determined as follows. The case of the triple frame shown in FIG. 11 will be described as an example. Similarly to the case of one frame, the level outside the outermost frame 61 is “0”. Since the area 64 between the outermost frame 61 and the second frame 62 needs to exceed at least one frame from the upper end, the pixel level in this area is "1". Similarly, the level of the area 65 between the second frame line 62 and the third frame line 63 is “2”, and the level of the area 66 inside the third frame line is “3”. Therefore, the regions 64 and 66 whose levels are odd are determined to be regions within the frame. This example is an example of a triple frame. However, when the number of frames further overlaps, the level is increased by one each time the next inner frame is entered, and the areas are alternately determined as the in-frame area and the out-of-frame area, respectively. As a result, a donut-shaped region indicated by a hatched portion and a region having a further isolated island in a hole of the donut can be easily designated by a frame line.

(In the case where the frame is not an ideal frame) In the above examples, the shape of the frame is an ideal closed curve.
In many cases, a noise such as "blurring" is added. In this processing, it is possible to stably detect some of these noises. Next, as an example of noise, "whiskers"
An example of a case where a “nest” or the like occurs will be described.

FIG. 12 shows a case where a “whisker” 72 occurs outside the frame line. As in the above example, the level of the area outside the frame line and the level of the area inside the frame line are “0” and “1”, respectively. In addition, the "beard" part of the border, the border and the "beard" of the border
73 surrounded by the raster at the lowest point (shown by a broken line in the figure)
Needs to exceed at least one frame line from the upper end of the image, so the level is "1". Therefore, the area 73 that is originally appropriate to be regarded as the outside area is also determined as the inside area. However, since such an area 73 is limited to a small area between the “whisker” and the frame line and above the lowest point of the “whisker”,
There is no practical problem with this determination error.

Further, as shown in FIG. 13, small holes (hereinafter, referred to as nests) 81 and 82 determined to be the ground may occur inside the frame line. This is caused by, for example, unevenness in writing a frame line. Note that, in FIG. 13, the hatched portion represents a frame line portion. In the small nest 81, the level inside the nest becomes “1” and is determined only as an area within the frame, but does not affect other parts. In addition, when the nest is extremely large, for example, when the lowermost part of the nest is lower than the uppermost part of the inner area of the frame like the nest 82, the area inside the frame and above the raster 83 at the lowermost part of the nest. The level of 85 is "2", and is determined to be out of the frame. But raster
The area below 83 is a route that does not cross the border line only once from the upper end
Since 84 exists, the level becomes “1”, and this portion is determined to be a region within the frame. Such nests are usually not large,
The area of the erroneous determination does not exceed the size of the nest, so there is no practical problem.

As described above, according to the in-frame detection processing of the present embodiment, there are the following advantages.

(1) Basically, if the frame is a normal frame, the in-frame area can be detected whether the frame is downwardly concave or multiple frames.

(2) Among various noises generated in writing the frame line,
Beard, nest, etc. are hardly affected.

(3) Regarding the break of the frame line, if it is below the frame line, the area within the frame is correctly detected.

That is, it is possible to cope with a complicated frame line such as a concave frame or a multiple frame, and is resistant to noise in writing the frame line. It is difficult for a human to draw an outline, and to create an ideal closed curve (without cuts and whiskers) by precisely matching the start and end points. The in-frame detection processing of the present embodiment is resistant to noise such as small whiskers and nests if care is taken so as not to cause breakage of the frame line. Is improved.

Next, a specific configuration of the in-frame area detection device according to the present embodiment will be described. FIG. 15 is an example of a circuit configuration of the in-frame area detecting device that realizes the algorithm shown in FIGS. 2 and 3, and FIG. 16 is a time chart of each part for explaining the operation.

15, a frame signal (P) 101 from the frame signal generating circuit 2 in FIG. 1 is input to a terminal 100. This frame signal 1
01 is obtained by determining from the input image signal obtained by raster scanning one raster at a time from the upper left to the right in the image input unit 1 in FIG. Signal is “1” in the frame and “0” in the other
Is a binary signal. The frame signal 101 is input in synchronization with a clock signal (CLK) 102.

Negation of the frame signal 101 and the frame signal (P0) 103 one raster before.
Is obtained by the gate circuit 104, and the result is added to the level signal (L0) 106 one raster before by the adder 105. The output signal 107 of the adder 105 corresponds to the provisional level signal calculated by the first main step of the in-frame detection algorithm described with reference to FIGS.

Next, an exclusive OR of the frame signal 101 and a signal obtained by delaying the frame signal 101 by one pixel by the latch 108 is obtained by the gate circuit 109. Since the output signal 110 of the gate circuit 109 indicates the end of the connection area, it is hereinafter referred to as an area end signal. The counter 1 counts up in synchronization with the clock signal 102.
The coordinate signal 112, which is the output of 10, is latched by the latch circuit 113 when the area end signal 110 is "1", and is output as the area position signal 114. The region end signal 110 indicates a boundary from the frame to the ground or from the ground to the frame, and the region position signal 114 indicates the position (x coordinate) of the boundary.

On the other hand, the area minimum value calculation circuit 115 calculates the minimum value of the level signal 107 in the connected area, that is, the representative level signal. The area minimum value calculation circuit 115 is specifically composed of a minimum value circuit 116 and a latch circuit 117. Each time the area end signal 110 becomes “1”, the latch circuit 117 sets the level maximum value (in this embodiment, “15”) is set. The smaller of the output of the latch circuit 117 and the level signal 107 is detected by the minimum value circuit 116, latched and updated by the latch circuit 117, and the minimum value of the level signal in the connection region is calculated. Output signal 118 of area minimum value calculation circuit 115
Corresponds to the representative level signal calculated in the second main step of the in-frame detection algorithm described with reference to FIGS. 1 and 2.

This representative level signal 118 is synchronized with the area end signal 110, and is a signal for each connected area. In order to convert the signal in the unit of connected area into a signal in the unit of pixel, the level signal 118 is written into the memory unit 119 for one raster. That is, each time the area end signal 110 becomes “1”, the frame signal 101, the representative level signal 118, and the area position signal are stored in the memory unit 119.
114 is written. Therefore, the number of words to be written to the memory unit 119 is the number of areas of one raster. The information stored in the memory unit 119 is read out as a pixel unit signal while referring to the area position signal 114 at the time of the next raster processing. Specifically, the area position signal 124 read from the memory unit 119
And the coordinate signal 112 from the counter 111 is compared by a comparator 125, and the address of the RAM 120 in the memory unit 119 is incremented by one each time they become equal. Thus, frame signal 103 and level signal 106 read from memory unit 119
Is a signal in pixel units. These frame signal 103 and level signal 106 represent information of one raster earlier and are used for calculation of the next raster. Here, the memory unit 119 needs two rasters, but in the present embodiment, two address counters 121 and 122 for reading and writing are used, and the address is switched in a time-division manner by the switch 123 so that one raster is used. RAM1
I'm done with 20.

Finally, the gate circuit 126 performs a logical sum of the least significant bit of the level signal 106 read from the memory unit 119 and the frame signal 103, and outputs the logical sum of the inside / outside frame determination signal 127.
Is output as In the in-frame / out-frame determination signal 127, "1" indicates the inside of the frame, and "0" indicates the outside of the frame. In the present embodiment, the frame line is regarded as being outside the region. However, by changing the gate circuit 126, the frame line portion can be regarded as being within the region.

As described above, the in-frame area detection device according to the present invention can be realized with a simple configuration. In particular, by storing a representative level signal of one raster in the memory unit 119 in a connected area unit, that is, in a run format, one raster delay for decoding a run and referring to the next raster can be shared. There is an advantage that the capacity of the device can be small.

FIG. 20 is a block diagram of an in-frame area detecting apparatus according to a second embodiment of the present invention. A representative level signal correction circuit 6 is inserted after the representative level signal calculation circuit 4. In this embodiment, in other words, the second main step for determining the level in the raster, that is, the calculation of the representative level signal in the first embodiment is modified as in the following <Rule 4>.

<Rule 4> When frame pixels or ground pixels are adjacent to each other in the same raster and their level signals are different, the smaller of the level signals is set as the representative level signal of those pixels. . Up to now, the calculation is the same as the calculation of the representative level signal in the first embodiment. In this embodiment, as a process of correcting the representative level signal, if the frame pixel and the ground pixel are adjacent to each other and the level signal of the frame pixel is larger than the level signal of the ground pixel, the level signal of the ground pixel is changed to the level signal of the frame pixel. And If the frame pixel and the ground pixel are adjacent to each other and the level signal of the ground pixel is larger than the level signal +1 of the frame pixel, the level signal +1 of the frame pixel is set as the level signal of the ground pixel (see FIG. 21).

 The above rule can be expressed by the following equation.

if [(P (x, y) = P (x + 1, y)] and [L (x, y) ≠ L (x + 1, y)] then L (x, y) = L (x + 1, y) = Min [ L (x, y), L (x + 1, y)] if [P (x, y) = 1] and [P (x + 1, y) = 0] and [L (x, y)> L (x + 1, y )] Then L (x, y) = L (x + 1, y) if [P (x, y) = 1] and [P (x + 1, y) = 0] and [L (x, y) +1 <L ( x + 1, y)] then L (x + 1, y) = L (x, y) +1 if [P (x, y) = 0] and [P (x + 1, y) = 1] and [L (x, y) <L (x + 1, y)] then L (x + 1, y) = L (x, y) if [P (x, y) = 0] and [P (x + 1, y) = 1] and [L (x, y) y)> L (x + 1, y) +1] then L (x, y) = L (x + 1, y) +1 The above processing is repeated until there is no pixel set satisfying the above condition in the raster. In the
The judgment operation in the inside / outside frame judgment circuit 7 in FIG.
This is slightly different from that of the in-frame / out-of-frame determination circuit 5 of FIG. The level signal of the pixel P determined according to the present embodiment indicates the downward distance from the upper end of the image to the pixel P, and the definition of the path following the downward direction is based on the definition 2 below.

Definition 2: A downward path is a path that follows an adjacent pixel in the horizontal direction or downward direction.

That is, in the first embodiment, a path in the left-right direction is not allowed between different kinds of pixels, but in this embodiment, this restriction is eliminated. Therefore, even in the case of a frame figure that is difficult to determine accurately in the first embodiment, correct determination can be made. This will be described with reference to the example of FIG. In this example, the border of the "beard" is extremely large.
Both 92 and the tip 93 of the "whisker" extend to the point where it narrows below the border. In this case, according to the first embodiment, since the route from the upper end to the area inside the frame must necessarily cross the frame at least twice, the nesting levels inside this frame are all "2", It is determined to be out of the frame. However, it is considered rare that such a frame line usually occurs, and there is no serious problem in practical use. On the other hand, according to the second embodiment, for example, an area below the raster 94 in the frame line is determined to be inside the frame. For this reason, it is possible to detect the in-frame area that is more resistant to noise and the like than in the first embodiment.

In the first and second embodiments, only one pixel of the entire raster immediately above the pixel of the current raster shown in FIG. 17A is defined as the adjacent relationship in the upward direction. As an example of the relationship, two pixels right above and diagonally above right (or diagonally above left) as shown in FIG. 4B, and three pixels right above, diagonally above right and diagonally above left as shown in FIG. The concept of a pixel may be used. In this case, the algorithm for calculating the level signal in the first main step is as follows.
The provisional level signal of the pixel may be calculated according to Rule 1>, and the minimum value thereof may be used as the representative level signal of the pixel. In particular, FIG. 17 (b) has a feature that when the frame area and the ground area intersect, one of the areas is always adjacent to the other.

In addition, the present invention can be implemented with various modifications.

[Effects of the Invention] According to the present invention, even in a region surrounded by a double or more multiple frame line or a lower concave frame line, the lower concave portion or the inside of the double frame can be determined to be outside the frame. It is also possible to stably determine noise such as "whiskers" generated by handwriting. Further, in the present invention, when calculating the minimum value of the nest level in the main scanning direction, that is, the representative level signal,
The conversion into the connected area unit can be used, and it can be realized with a simple circuit. Therefore, it is possible to provide an in-frame detection device that has less restrictions on the user in writing the frame line and has a small circuit scale.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an in-frame area detecting apparatus according to a first embodiment of the present invention, FIG. 2 is a diagram showing an outline of an in-frame detection algorithm in the embodiment, and FIG. FIG. 2 is a diagram showing a part of the algorithm of FIG. 2 in detail, FIG. 4 is a diagram for explaining setting of coordinate axes on an image, FIG. 5 is a diagram for explaining contents of a first main step of the algorithm, FIG. 6 is a diagram for explaining the contents of the second main step, FIGS. 7 and 8 are diagrams for explaining that the level signal and the lower distance are equivalent, and FIGS.
FIG. 14 is a diagram for explaining a specific example of the in-frame area detection according to the embodiment, FIG. 15 is a circuit diagram showing in detail a main part of the in-frame area detection device of FIG. 1, and FIG. FIG. 17 is a time chart for explaining the operation of FIG. 15, FIG. 17 is a diagram for explaining the definition of the upper and lower adjacent relationship of the pixels, FIG. 18 is a diagram for explaining the minimum over-frame route and the distance, and FIG. FIG. 20 is a diagram for explaining a lower minimum frame passing path and a lower distance, FIG. 20 is a block diagram showing a schematic configuration of an in-frame area detecting apparatus according to a second embodiment of the present invention, and FIG. FIG. 9 is a diagram for explaining a second main step in FIG. DESCRIPTION OF SYMBOLS 1 ... Image input part 2 ... Frame signal generation circuit 3 ... Tentative level signal calculation circuit 4 ... Representative level signal calculation circuit 5,7 ... In-frame / out-frame area determination circuit 6 ... Representative level signal correction circuit 101 ... Current raster frame signal 103... One raster previous frame signal 106... One raster previous level signal 107... Temporary level signal 118... Current raster representative level signal 127...

────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kazuhiko Higuchi 1 Toshiba Research Institute, Komukai, Kawasaki City, Kanagawa Prefecture (56) References JP-A-3-17783 (JP, A) JP Hei 3-17787 (JP, A) (58) Fields studied (Int. Cl. ⁶ , DB name) G06T 7/40

Claims (2)

(57) [Claims] 1. An in-frame area detecting method for detecting an in-frame area surrounded by a frame line on an image from an input image signal obtained by scanning an image including the frame line. A binary frame signal is generated for each pixel by determining whether or not the pixel is a frame portion. For a current pixel of the input image signal, a frame signal of a previous pixel adjacent to the current pixel, a level signal of the previous pixel, and A tentative level signal is calculated from the frame signal of the current pixel, a connected region in which the frame signal values are equal and continuous is detected for the current pixel column in which the tentative level signal is calculated, and a temporary region of each pixel in the connected region is detected. Calculating a level signal of a pixel in the connected area from a value of the level signal, and determining an inside or outside of the frame from the value of the level signal of the pixel in the connected area to detect the area within the frame. A method for detecting the area within the frame. 2. An in-frame area detection method for detecting an in-frame area surrounded by a frame on an image from an input image signal obtained by scanning an image including the frame. A binary frame signal is generated for each pixel by determining whether or not the pixel is a frame portion. For a current pixel of the input image signal, a frame signal of a previous pixel adjacent to the current pixel, a level signal of the previous pixel, and A tentative level signal is calculated from the frame signal of the current pixel, a connected region in which the frame signal values are equal and continuous is detected for the current pixel column in which the tentative level signal is calculated, and a temporary region of each pixel in the connected region is detected. Calculating the level signal of the pixel in the connected region from the value of the level signal, correcting the level signal of the pixel in the connected region using the relationship between the level signal of the pixel in the adjacent connected region and the frame signal; Judge the inside and outside of the frame from the corrected level signal value Frame area detecting method and detecting the frame region Te.