JPH0362676A - Picture processing unit - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image processing device for decoding encoded image information such as a facsimile machine.

[Conventional technology]

When image information is transmitted via a line using a facsimile machine, videophone, videotex, electronic publishing system, etc., the image information is first encoded on the sending side and then transmitted, and then decoded on the receiving side. A method is used to prevent signal deterioration in the transmission system.

Incidentally, in the past, the process was simply to transmit information, but in recent years, various processes (for example, image cutting, etc.) have been performed in the process up to image restoration.

[Problem to be solved by the invention]

However, in the conventional image processing device described above, when performing image cropping, image compression processing and image cropping processing are performed separately, and even when outputting a part of the compressed image, the entire screen is The method used was to perform extraction processing after restoring the image.

Therefore, even when a part of the image is obtained, a large-capacity image memory is required to decode the entire screen, and the decoding process requires a large amount of time.

SUMMARY OF THE INVENTION An object of the present invention is to provide an image processing apparatus that can solve the problems of the prior art and can perform decoding processing at high speed without increasing the memory size.

[Means to solve the problem]

In order to achieve the above object, the present invention performs image processing to obtain a restored image by performing interpolation processing on image data that encodes the density and distance of points where the difference in density of each pixel changes. The device specifies the horizontal and vertical start points and end points of the cropped image, starts restoration by detecting the start point, and interpolates and stores the difference between density change points if it is not a blank section. , processing means for terminating the restoration upon detection of the end point is provided.

[Effect]

According to the above means, the l-line is determined for the encoded image data based on the distance of the density change point, and the restoration of the change point is started from the time when the specified cutting start point is detected. In this restoration process, interpolation is performed if the interval between density changes is not a blank section. Therefore, conventionally, −
In contrast to decoding the entire image and then cutting out only the necessary parts, the decoding time can be shortened and the memory capacity can be reduced.

〔Example〕

FIG. 1 is a block diagram showing one embodiment of the present invention, and FIG. 2 is a block diagram showing the configuration of an encoding system corresponding to the decoding system of FIG. 1.

First, the configuration of the encoding processing system will be explained with reference to FIG.

In the image input device l, an image memory 2 and a density change point detection circuit 3 are connected in sequence, and a horizontal direction change point memory 4 and a vertical direction change point memory 5 are connected in parallel to the density change point detection circuit 3. . A thinning circuit 6 is connected to these memories 4 and 5, a Huffman encoder 7 is connected to the thinning circuit 6, and a storage device 8 is connected to store the results of the encoding process.

The encoding process in the above configuration will be explained with reference to the flowchart in FIG.

The image input device 1 converts images with shading, such as photographs, into CC.
Read using a sensor such as D (charge coupled device),
The read information is stored in the image memory 2. A density change point detection circuit 3 detects points at which the image density changes with respect to data stored in the image memory 2. With respect to this density change point detection data, the position of the concentration change point in the horizontal direction is stored in the horizontal direction change point memory 4, and the position of the vertical direction concentration change point is stored in the vertical direction change point memory 5 (step 3
1.32).

The thinning circuit 6 performs thinning processing on the horizontal density changing point and vertical density changing point information stored in the horizontal changing point memory 4 and the vertical changing point memory 5 (step 33). The Huffman encoder 7 performs Huffman encoding on the density and distance for the density change points resulting from this thinning process (steps 34 and 35). This encoding processing result is stored in the storage device 8.

Here, the encoding process will be explained in detail.

The above-described encoding process detects edge pixels and encodes their density and distance between edge pixels. Here, as shown in FIG. 4 (in the case of maximum thinning error Mε<threshold value) or FIG. 5 (in the case of maximum thinning error Mε≧threshold value), the density of each pixel in the horizontal direction is denoted by a, −
It is assumed that aN constitutes an l line.

First, a, which is the starting point of a line, is set as a first edge pixel, and it is determined whether the pixel a is an edge pixel. In the following, the thinning error is calculated assuming that a2 is not an edge pixel.

Let the density of the first edge pixel be al, and a2 be a.

Then, the second edge pixel when a+ is thinned out is (
a+++). Based on this information, the density bK (x<k≦j) of the pixels between the edges is calculated as follows.

bK=a,+ ((a+ +1 8t)/(j+
1-i) ) X (k-i)... (1) Comparing this with the original density a, the thinning error Δa,
is calculated as follows.

Δak=lbK−akl ---(2) Next,
The maximum value of the thinning error (hereinafter referred to as maximum thinning error) Mε in the interpolation section is as follows.

Mε=MAX (Δa,) (where ink≦j) If Mε is smaller than the threshold, it is determined that the pixel can be thinned out (that is, it is determined that it is not an edge pixel), and conversely, Mε
is larger than the threshold, it is determined that the pixel is an edge pixel. If it is determined that it is not an edge pixel, Mε is calculated using the next pixel (al+1) as a new thinning target, and a new assumption is determined. If it is determined that it is an edge pixel, it is determined as the second edge pixel, and after recording the density of the first edge pixel and the distance (j+1-i) between the edge pixels, a+ is set as the new first edge pixel. Then, similar processing is performed for subsequent pixels.

For the horizontal edge pixel information obtained in this way, an interval (hereinafter referred to as a blank interval) in which no horizontal interpolation is performed during restoration is created based on the vertical edge pixel information obtained in a similar manner. insert.

That is, the horizontal edge pixel NX+l shown in FIG.
, V and the next edge pixel NX+L, V which is separated by the distance
If there is a vertical edge pixel in the section between
The mark indicates a horizontal edge pixel, and the circle indicates a vertical edge pixel).

Furthermore, if there is no edge pixel in the vertical direction in the same section, the information in that section is determined by pixels that can be restored by vertical interpolation, and is set as a blank section in the horizontal direction. Furthermore, if several blank sections are consecutive, their distances are summed up to form one new blank section. FIG. 7 shows a state in which the thinning of edge pixels has been completed by inserting a blank section (in the figure, X marks are horizontal edge pixels, and Δ marks are blank sections).

Note that when encoding the state of the edge pixel shown in FIG. 7, conversion is performed using the format shown in FIG. 8.

Basically, I sections are represented by the density and distance of edge pixels. For a blank section, first, the distance is set to 0 (a distance of 0 is impossible in a normal edge section) to indicate that it is a blank section, and then the distance of the blank section is indicated. That is, the normal section is indicated by (density 10 distance + density), and the blank section is indicated by (density 〇 + distance 10 density).

Next, this data is Huffman encoded. To improve encoding efficiency, edge pixel information is separated into density and distance. For density, the difference from the previous edge pixel is encoded, and for distance, the value as is is encoded using another Huffman table.

Next, the configuration of the embodiment (decoding processing system) shown in FIG. 1 will be explained.

The storage device 8 includes a Huffman decoder 9 and a count circuit 10.
, a change point restoration circuit 11, and a change point memory 12 are connected in sequence. A horizontal interpolation circuit 13 and an interpolated status memory 14 are connected in parallel to this change point memory 12. A vertical interpolation circuit 15 is connected to the interpolated status memory 14, and an image memory 16 is connected to the horizontal interpolation circuit 13. Also, image memory 1
6 is connected to a display device 17.

Next, the operation of the configuration shown in FIG. 1 will be explained with reference to FIG. 9.

The Huffman decoder 9 reads the encoded data from the storage device 8 and decodes the density and distance of the density change point of the image information (steps 91, 92). With respect to this decoded data, a count circuit 10 counts pixels and lines in preparation for extraction processing to be described later. Further, a change point restoration circuit 11 restores the density change points based on the density and distance, and stores the density change points of the cut out image in the change point memory 11.
2.

The horizontal interpolation circuit 1 uses the data in the change point memory 12.
3 interpolates horizontally between the change points (step 93
). Further, the interpolated status memory 14 stores data with a status indicating that interpolation has been completed in the horizontal direction. Based on this interpolated status, vertical interpolation is performed by the vertical interpolation circuit 15 (step 94
). The image information restored by horizontal and vertical interpolation processing in this manner is stored in the image memory 16, read out as necessary, and displayed on the display device 17.

Next, details of the decoding process will be explained with reference to FIGS. 9, 10, and 11.

Horizontal edge pixel information processed by Huffman encoding is
Decoded to obtain horizontal edge pixel information. The density of the image is decoded based on this information, but at this point the 7th
An image similar to the one shown in the figure is obtained.

Here, edge pixel sections are determined and interpolated in the horizontal direction as shown below. Let the first edge pixel be STx,v, and the second edge pixel separated by a distance from here be ENX+LV
Then, the edge pixel Hx+s,v which is separated by N from the first edge pixel STx,v is as follows.

HX+N v = S Tx, v + ((E NX
+L, v-8Txv)/L) Furthermore, if the edge pixel section is a blank section, interpolation processing for that section is not performed, and the corresponding section in the status memory is recorded as unrestored. FIG. 1O shows the contents of the image memory in this state (in the figure, the x marks are horizontal edge pixels, and the * marks are horizontal interpolation pixels).

Next, perform vertical interpolation. First, the data stored in the memory is scanned in the vertical direction to detect a restored pixel (this pixel is referred to as the first edge pixel STx,v), and the scanning is further performed to detect the next restored pixel. (This is set as the second edge pixel ENX+L, V). Letting the distance between them be L, HX+NV, which is separated by N from STx,v, is determined by the following equation and interpolated in the vertical direction.

Hx, V+N = S Tx, v + ((E N
x, V + L - 8Tx, v) / L) Restore. This state is shown in FIG. 11 (in the figure, marks are vertical interpolation pixels, * marks are horizontal interpolation pixels, and x marks are horizontal edge pixels), and in this state, the image restoration process ends.

Note that when scaling an image during restoration, the size can be freely changed by multiplying the distance by the desired magnification.

Next, a method of performing restoration processing while cutting out images will be described.

First, specify the horizontal and vertical starting and ending points of the image to be cut out. Next, the distance of the density change point obtained by the Huffman decoder 9 is added to a pixel counter (not shown) by a counting circuit 10.

If the total matches the horizontal size of the original image, this is determined to be one line, a line counter (not shown) is added, and the pixel counter is cleared at the same time. The above process is repeated until the line counter matches the vertical start point, and when it matches, the process is repeated until the pixel counter matches or exceeds the vertical start point. If this starting point is exceeded, the changing point restoration circuit II starts restoring the changing point from the density changing point at the time when the sizes match, going back to the previous density changing point.

At this time, if the density change point is not a blank section, the horizontal interpolation circuit 13 interpolates the change point section, stores pixels corresponding to the cut-out image in the image memory 16, and further stores interpolated information in the interpolated status memory 14. remember. The pixel counter continues to be incremented even during the restoration of the change point, and when the value exceeds the cutout end point, the restoration of that line is completed, the line counter is added, and the pixel counter is cleared at the same time. This process is repeated until the line counter counts the end point of the vertical cutout, and then the vertical interpolation circuit 15 restores unrestored pixels, and all processes end.

〔Effect of the invention〕

As is clear from the above, according to the present invention, an image processing apparatus obtains a restored image by performing interpolation processing on image data in which the density and distance of points where the difference in density of each pixel changes is encoded. In this step, the horizontal and vertical start points and end points of the cropped image are specified, restoration is started by detecting the start point, and if the density change points are not blank sections, interpolation and storage are performed between the change points, and Since a processing means for terminating the restoration upon detection of the end point is provided, less memory is required and decoding time can be increased.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a block diagram showing the configuration of an encoding system corresponding to the decoding system in FIG. 1, and FIG. 3 is a flowchart showing the encoding process. Figures 4 and 5 are explanatory diagrams showing the calculation of thinning errors;
The figure is an explanatory diagram showing the arrangement of horizontal and vertical edge pixels, Fig. 7 is an explanatory diagram showing the edge pixel arrangement after thinning, Fig. 8 is an explanatory diagram showing the format of edge pixel information, and Fig. 9 is the same as Fig. 1. FIG. 1O is an explanatory diagram showing a process after horizontal interpolation, and FIG. 11 is an explanatory diagram showing a pixel arrangement in a restored state. DESCRIPTION OF SYMBOLS 1... Image input device, 2... Image memory, 3... Density change point detection circuit, 4... Horizontal direction change point memory, 5...
・Vertical direction change point memory, 6... thinning circuit, 7...
- Huffman encoder, 8... Storage device, 9... Huffman decoder, 10... Count circuit, 11... Change point restoration circuit, 12... Change point memory, 13... Horizontal interpolation circuit, 14... Interpolated status memory, ■
5... Vertical interpolation circuit, 16... Image memory, 1
7...Display device. Figure Figure Figure 3 Figure Figure Figure Figure 3 Figure Figure Figure Figure Figure t j aj◆1 Figure Figure Figure 8 77 Nokl Pavilion

Claims (1)

[Claims] In an image processing device that obtains a restored image by performing interpolation processing on image data that encodes the density and distance of points where the difference in density of each pixel changes, each horizontal and vertical starting point of a cut-out image is used. , processing means that specifies an end point, starts restoration by detecting the start point, interpolates and stores the difference between density change points when the interval between density change points is not a blank section, and ends restoration by detecting the end point. An image processing device characterized by being provided with.