JPH099292A - Data structure considering sparseness of video information in dynamic image - Google Patents

Abstract

PURPOSE: To remarkably reduce the amount of data by making only information of color different from that of a fundamental picture in to data when the information of color of a target picture in a dynamic image is made into the data. CONSTITUTION: The making of the information of color of the target picture in the dynamic image into the data is performed by expressing only the information of color different from that of the fundamental picture as the data, and employing data structure considering the sparseness of picture information. In the expression of each data item, the kind and position (row number and column number) of the color of each picture element are used. The expression is performed by unifying the picture element at every row of the picture element, every column of the same, and every color of the same, etc., appropriately. The information of the color is expressed by separating into three primary colors.

Description

DETAILED DESCRIPTION OF THE INVENTION In a moving image, information on pixels having the same color as the basic screen is omitted, and only the information on pixels having a different color from the basic screen is taken into consideration for sparseness. Data structure. At that time, the color of the pixel, or
Sorting may be performed at pixel positions (that is, row numbers and column numbers), and some pixels may be sorted. This makes it possible to significantly reduce the amount of data relating to moving images. A moving image requires a large number of screens, but for the image of each screen, it is necessary to express, as data, what color all the pixels on the display have. There are millions of pixels and tens of millions of colors, and more than tens of screens of information are needed per second.
It becomes a huge amount of data. In the past, when expressing information about moving images as data, each screen was configured as data that had nothing to do with each other, even if the image did not change very rapidly and each screen hardly changed. . In the present invention, when each screen does not change much in a moving image, information on pixels having the same color (same brightness, saturation, etc.) as the basic screen is omitted, and only information on pixels having different colors is displayed. This is a data structure that takes sparseness into consideration. When constructing this data, sort by pixel color or pixel position (that is, row number and column number) etc. without dividing into individual pixels, and specify a pixel range of the same property. You may. This data structure makes it possible to significantly reduce the amount of data. The screen used for the following explanation is 2 as shown in FIG.
Although it has 0 rows × 20 columns of pixels, it is actually the number of pixels of the display for displaying the screen. Further, the data representing the color of the pixel is represented by a lower case alphabet. In FIG. 3, it is assumed that information about a screen that is the basis of a moving image (hereinafter, this is referred to as “basic screen”) is obtained. The information of the screen to be expressed as data (hereinafter referred to as "target screen") is shown in FIG. Normally, the "basic screen" may be the screen one before the "target screen". In the "Target screen", select "Pixel of different color" from "Basic screen"
It is shown in FIG. When the color of each pixel shown in FIG. 1 is expressed as data in consideration of its sparsity, it becomes as shown in FIG. 2 as an example. In the example of the “target screen” shown in FIG. 4, the data reduction rate by considering the sparsity is as follows. The row number and the column number of each pixel are represented by 2 bytes, and the color is represented by 3 bytes (or 4 bytes). In FIG. 4, 3 bytes × 20 × 20 = 1200 bytes or 4 bytes × 20 × 20 = 1600 bytes. As shown in FIG. 1, since 22 pixels are different, (2 + 2 + 3) bytes × 22 = 154 bytes or (2 + 2 + 4) bytes × 22 = 176 bytes, which is about 13% using the data structure of FIG. (Or 11%). The information of the "pixels of different colors" shown in FIG. 2 is a data item which itself is different, and therefore may be collected for each data item. That is, each data item requires the type of color of each pixel and the position (that is, the row number and column number of the pixel).
Sorting is performed on either the row number or the column number, and all data items are grouped into some data items and expressed as a plurality of data groups. The following three data structures are mainly conceivable for this way of summarizing.

[1] Method of grouping pixels by row This is a method of grouping, for each row, the number of columns to the number of colors as one unit. FIG. 5 shows an example of the data structure according to this way of grouping.

[2] Method of grouping pixels by column This is a method of grouping, for each column, the number of rows and the number of colors as one unit. FIG. 6 shows an example of the data structure according to this way of grouping.

[3] Method of grouping pixels by color It is a method of grouping each color by assuming that the same color is applied to the rows from the rows to the rows to the columns. 7 and 8 show examples of the data structure according to this organization method. FIG. 7 shows the way of grouping the row numbers first, and FIG. 8 shows the way of grouping the column numbers first. Alternatively, a method of dividing a set of pixels into some figures, for example, rectangles (or squares) and showing two non-adjacent corners of the four corners of each rectangle is also conceivable. FIG. 9 shows an example of the data structure according to this way of grouping. The color information may be expressed as data by dividing it into three primary colors. In this case, the number of color data elements is 3 instead of 1, and the data may be sorted by each element. For example, the three primary colors are the primary color 1, the primary color 2, and the primary color 3, and the types of colors related to the respective primary colors are primary color A, primary color B, and primary color C, respectively.
Is expressed as {line number, column number, (primary color A, primary color B, primary color C)}. Alternatively, by providing a "target screen" for each primary color, considering three screens, (row number, column number, primary color A) (row number, column number, primary color B) (row number, column number, primary color As C), similar sorting will be performed.

[Brief description of drawings]

[FIG. 1] In the “target screen”, the “basic screen” and “pixels of different colors” (display pixels are set to 20 × 20,
The alphabet represents color. )

[Fig. 2] Data structure of "target screen" in consideration of sparsity

FIG. 3 is a basic screen of a moving image (that is, “basic screen”)

[Fig. 4] "Target screen" of moving image

FIG. 5 is a data structure in which “pixels of different colors” are grouped in rows on the “target screen”.

FIG. 6 is a data structure in which “pixels of different colors” are collected for each column in the “target screen”.

[FIG. 7] Data structure in which “pixels of different colors” are grouped by color in the “target screen” (when line numbers are described first)

FIG. 8 is a data structure in which “pixels of different colors” are grouped for each color in the “target screen” (when the column number is described first).

FIG. 9 is a data structure in which “pixels of different colors” are grouped for each color in the “target screen” (when described for each rectangular block).

Claims (1)

[Claims] 1. When expressing what color each pixel is for each screen as information relating to a video in a moving image,
A data structure that considers sparsity, omitting data related to "pixels having the same color" as the basic screen and representing only data related to "pixels having a different color". In this data structure, data about "pixels of different colors"
Sorting may be performed on either the pixel color or the row number or the column number of the pixel, and some pixels may be sorted.