JPS62120181A - Adaptive encoding method for digital multivalued picture - Google Patents

Abstract

PURPOSE:To perform an approximation with less error by the reduced number of representative blocks by forming a zero average picture element block restricting a statistical dispersion as an approximation object block. CONSTITUTION:An average value A in a picture element block is calculated in a zero average picture element block producing part 2 with respect to picture data read by a picture reading part 1 and the zero average picture element block is formed. A picture element value in a representative block is adaptively selected from the population of the picture element value in a picture element value memory 3 in a picture element value selecting part 4 in the representative block. An encoding part 5 selects the representative block and encodes the type of the representative block and the type of the picture element value of the population selected as the picture element value of the representative block. An approximation picture reproducing part 9 reproduces an approximation picture element block from an approximation zero average block reproduced in an approximation zero average picture element block reproducing part 8 and the average value A decoded in a decoding part 6. A picture display part 10 arranges one frame of the reproduced approximation picture block and displays.

Description

[Detailed description of the invention] (Industrial application field) The present invention relates to a video conference system and a monitoring system.

The present invention also relates to an adaptive encoding method for digital multilevel images used in still image transmission devices and the like.

(Prior Art) Conventionally, this type of encoding method selects an approximate representative block from a set of representative blocks prepared in advance for a pixel block, and then assigns each pixel block that has been allocated in advance to each representative block. A binary code representing the type of representative block is transmitted.

The decoding side selects a representative block corresponding to the transmitted binary code from a table of representative block sets prepared in advance and uses it as a reproduced block.

FIG. 9 is a block diagram of a conventional encoding method.

In the figure, reference numeral 11 denotes an image reading section, which divides the image read into pixel blocks. 12 is a memory pattern containing a set of representative blocks predetermined in the table of representative processor 1-. Reference numeral 13 denotes an approximate button selection section which selects from among the memory buttons 12 a representative block that is most similar to the pixel blocks divided by the image reading section 11.

Reference numeral 14 denotes an encoding unit which transmits the binary code previously allocated to the representative block selected by the approximate button selection unit 13. 15 is a decoding unit which receives the binary code transmitted from the encoding unit 14 and selects the binary code received by the decoding unit 15 from among the memory patterns 16 containing the same representative block set as the memory pattern 12. The approximate image restoring unit 17 restores the representative block corresponding to the code. After the codes are decoded for all pixel patterns within the screen, the image is displayed on the image display section 18.

(Reference: Japanese Patent Application Laid-open No. 55-150079) (Problems to be Solved by the Invention) In the conventional encoding method described above, since the buttons of all representative blocks are stored in the memory pattern, As the number of block types increases, a larger memory capacity is required, and in general, the approximation button selection section must first compare all sets of input pixel blocks and representative blocks before determining which representative block is approximated. It takes a lot of time to select one representative block because it cannot be determined whether it is a representative block or not. Furthermore, in the conventional method, the set of one representative block was fixed for any input pixel block, so when a certain input Regarding images, there was a drawback that the image quality was significantly degraded when displayed on the image display section.

An object of the present invention is to eliminate the drawbacks of the conventional technology, to be able to select a representative block with a small memory capacity, and to select a representative block with a very small amount of calculation, and to adaptively change a set of representative blocks according to the properties of an input image. An object of the present invention is to provide a method for adaptively encoding digital multilevel images.

(Means for Solving the Problems) In order to achieve the above object, the present invention provides a zero average value generated by subtracting the average value A of pixel values in an input pixel block from all pixel values in the input pixel block. A representative block that approximates the pixel block is selected, and the number of types of pixel values appearing in the representative block K (that is, only types of pixel values appear in u1 to ug) and the number of occurrences of each pixel value N8- N K (N l is the number of appearances of ul (i
= 1 to K), Σ N1 = number of pixels in a block) are fixed, a set of representative blocks is generated by rearranging these pixel values as much as possible, and the types of pixel values U, ~uK that appear are is a population of pixel values α1 to α, (P
``2K)''.

(Function) According to the above configuration, by generating a zero-average pixel block with suppressed statistical variation as the approximation target block, approximation with fewer errors can be performed using a smaller number of representative blocks, and one representative block can be approximated with less error. Ni sets of u
By simply rearranging the pixel values of i (i=l~K), the amount of calculation when selecting a representative block that approximates the zero-average pixel block can be extremely reduced, and furthermore, Memory for memory is no longer required, and the pixel values of the representative block can be adaptively selected from a pre-prepared population according to the zero-average pixel block, so calculations can be performed with less memory capacity. It is possible to obtain a reproduced image with little deterioration, which is not dependent on the statistical properties of the input image line.

(Example) An example of the present invention will be described based on FIGS. 1 to 8.

FIG. 1 is a block diagram of a method for adaptively encoding a digital multilevel image according to the present invention. In the figure, 1 is an image reading unit, and a zero-average pixel block generation unit 2 calculates the average value A within the pixel block for the image data read by the image reading unit 1, and generates a zero-average pixel block. . 3 is a pixel value memory, and the pixel values u1 to uK in the representative block are the population α□ to α of the pixel values in this double value memory 3,
The pixel values in the representative block are adaptively selected from among them by the pixel value selection unit 4 in the representative block. 5 is an encoding unit that selects a representative block, encodes the type of the representative block and the type of pixel value of the population selected as pixel values u8 to uK of the representative block, and a zero-average pixel block generator 2 The calculated average value is also transmitted. 6 is a decoding unit which decodes the code transmitted from the encoding unit 5, and by knowing the pixel values of the population selected as ul or u (from the pixel value memory 7), an approximate equal mean pixel block reproduction unit 8 performs approximation. A zero-average pixel block is reproduced. An approximate pixel block is reproduced by an approximate image reproducing unit 9 based on the approximate zero-average block reproduced by the approximate equal-average pixel block reproducing unit 8 and the average value A decoded by the decoding unit 6. The approximate image block reproduced by the approximate image reproducing unit 9 is
By arranging one frame, the reproduced image can be displayed on the image display section 10.

Next, the operation of the above embodiment will be explained. FIG. 2 is an explanatory diagram showing the principle of the present invention, showing an original multilevel image to be stored or transmitted.1. The pixel level of the original image is 8 levels (3 bits), and the size of the pixel block is 3×
It is set at 3. At this time, the pattern that may be expressed in the pixel block size is 8' = 134,217,
There are 728 types.

FIG. 3 shows part of a typical block set in this case. The representative block is composed of three types of pixel values, ul to u3, and the number of occurrences of each is N1 =
2, N2=5, N□=2. At this time, there are 756 types of representative blocks.

FIG. 4 shows an example of the population α1 to α of pixel values within a block. Let P=4, α, 20.

α2=1. Mu, which is defined as α3=2, α, ;3, is adaptively selected from among the populations α1 to α4.

Assuming that the upper left block in FIG. 2 is the input pixel block, FIG. 5 shows the input pixel block. Further, the zero-average pixel block and the average value A generated from this pixel block are shown in FIG.

For the zero-average pixel block shown in the same figure, pixel values u1 to U are adaptively set in the representative block as the pixel value α of the population.
Select from □ or α4.

Here, as an example of selection, we will select the absolute value of the two largest averages of the pixel values of the zero-average pixel block, and the pixel (on the number line) that is closest to the average of the two smallest averages. Select the value α+sin from α or α4, u=αwin, u,=:01u3=−αl1l
Defined as in. In the case of the example in Figure 6, αrnin = α
,=3. Therefore, U = -3゜u2 = 0, u, =
It becomes 3. Also, taking as an example a representative block that approximates a certain zero-average pixel block, the pixel values of the pixels with the largest pixel values 1 and 2 from the largest within the zero-average pixel block are U□
, then change the pixel values of the pixels with the fifth largest pixel value to u2, and change the pixel values of the remaining two pixels to U
.

It stipulates that it has been changed to A representative block for the zero-average pixel block in FIG. 6 is shown in FIG. The type of representative block generated in this way, information about which pixel value was selected as ul or U from among the pixel values of the population, and the average value A are transmitted. On the decoding side, an approximate image is reproduced by performing an operation opposite to that on the code side using the code of the type of representative block transmitted above, the information on the selected pixel value, and the average value A. An approximate image reproduced in this manner is shown in FIG.

In this way, according to the above embodiment, the representative block 756
Information on which pixel value was selected among the types (10 bits), information on which pixel value of the 14 types α1 to α was selected as Ul and u3 (2 bits), information on the average value A within the block It is only necessary to transmit a total of 15 bits (3 bits), which is 27 bits compared to the method of transmitting without any processing (9 pixels per block, 3 pixels per pixel).
A compression ratio of /15=1.8 has been achieved. In this example, the block size was set to 3 x 3 for ease of explanation, but by increasing the block size or reducing the number of pixel values in the population, visual image quality can be maintained without deteriorating much. , a larger compression ratio can be obtained.

(Effects of the Invention) According to the present invention, it is possible to transmit and store images with less deterioration in human visual perception with a small number of codes, and by creating a zero-average Hiro block with suppressed statistical variation as an approximation target block, Approximation with small errors can be performed using a small number of representative blocks, and the set of representative blocks can be
By simply rearranging the pixel values of ul (i=1 to k), the amount of calculation when selecting a representative block that approximates a block with zero-average pixels can be greatly reduced. Since no memory is required to store the pattern of one representative block, and the pixel values of the representative block can be adaptively changed according to the zero-average pixel block, better approximation can be achieved, which has practical effects. is a person.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an apparatus for carrying out an adaptive encoding method for digital multilevel images according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing the principle of the present invention, and FIG. 3 is a definition of the representative block. FIG. 4 is an explanatory diagram showing an example of the definition of the population of the same pixel value, FIG. 5 is an explanatory diagram showing an example of the same pixel block, and FIG. 6 is the zero mean of FIG. 5. FIG. 7 is an explanatory diagram showing an approximate representative block in FIG. 6. FIG. 8 is an explanatory diagram showing an example of an approximate reproduced image. FIG. 9 is an explanatory diagram showing an example of an approximate reproduced image. FIG. 9 is an explanatory diagram showing an example of an approximate reproduced image. 1 is a block diagram of an apparatus implementing the encoding method; FIG. 1... Image reading unit, 2... Zero average pixel block generation unit, 3... Pixel value memory, 4... Pixel value selection unit in representative block, 5... Encoding unit, 6.
...Decoding unit, 7...Pixel value memory, 8...
Approximate reproduction image block reproduction section, 9... Approximate image reproduction section, 10... Image display section. Patent applicant Matsushita Electric Industrial Co., Ltd. Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 1.2

Claims (1)

[Claims] A digital multivalued image is decomposed into pixel blocks composed of a plurality of pixels, the average value of the pixel values in the pixel block is calculated, and from all the pixel values in the pixel block,
generating a zero-average pixel block by subtracting the average value, and a number of representative blocks less than the number of all pixel patterns that may occur in the average pixel block;
The set of representative blocks in the method of approximating the average pixel block is generated by all possible rearrangements of these pixels, with the number of types of pixel values appearing in the representative block K and the number of occurrences of each pixel value being fixed. In this method, K kinds of appearing pixel values in the representative block are defined as a predetermined group of pixel values (α_1 to an adaptive code for a digital multivalued image, comprising: a step of adaptively selecting the representative block from α_P); and a step of selecting the representative block using the adaptively selected pixel values in the representative block. method.