JPH0126591B2 - - Google Patents

Description

[Detailed description of the invention]

INDUSTRIAL APPLICATION FIELD The present invention relates to a method for encoding multivalued image information. Configuration of conventional example and its problems When dividing a halftone image into pixels by scanning reading, quantizing the amplitude (brightness) of these pixels, and reducing the number of codes, the reading brightness range of pixels is divided into pixels at equal intervals. Divide by level, and give each level a level value that represents the level range. Then, it is determined which of the above levels the total value of the read luminance value of the pixel and the residual luminance signal of the previously processed pixel is at, and the value of the pixel is expressed as a value representative of that level. There is a conventional method of reducing the number of codes when a multivalued image is quantized by adding the difference between the total value and the representative value to a pixel to be processed after the pixel in question. However, this method has the disadvantage that because the levels are divided at equal intervals, the degree of image quality deterioration is not visually the same between high-brightness image parts and low-brightness image parts, and the deterioration is significant in dark parts. has. Purpose of the Invention The present invention makes it possible to reduce the number of bits per pixel with almost no deterioration in image quality when encoding brightness information of multivalued images, and the degree of deterioration in image quality can be reduced when encoding brightness information of multilevel images. The object of the present invention is to provide a method for quantizing multivalued image information such that both image regions and low image regions are visually almost the same. Structure of the Invention The present invention divides the range of the read luminance signal that a pixel can take into n regions, at least two of which are unequal, sets a region representative value representing each region in each region, and sets the luminance value. Determine in which area the brightness signal obtained by adding the residual brightness signal of the pixel that has already been quantized to the read brightness value of the pixel to be quantized (referred to as the relevant pixel) is located, and then quantize the relevant pixel. value as a representative value of the area, and the difference between the representative value and the luminance signal is added to the read luminance value of one or more pixels processed after the pixel in question, and the representative value representative of the area mentioned above is determined by the quantum The above objective is achieved by expressing the information using converted codes. Furthermore, as a method of dividing the above-mentioned n areas with unequal intervals, the intervals can be made finer in areas with low brightness, and the intervals can be increased in areas with high brightness. When it is proportional to , an image with less loss of visual naturalness can be obtained with a small number of codes. DESCRIPTION OF EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram of a system using the present invention. The above-mentioned addition process of the residual luminance signal will be specifically explained below. Four consecutive pixels a, b, c, in the horizontal direction of the screen
When the luminance signal value of d is set to "135" and the luminance signal area is set as shown in Table a below,

[Table] If no processing is performed, the luminance signals of these four pixels will be encoded with a representative value of "128", and the image will deteriorate as the luminance of all four pixels is "7" lower than the actual image. .

[Table] Therefore, in the present invention, as shown in Table b, the difference between the luminance signal value of pixel a and the representative value (residual luminance signal) = "7" is added to the luminance signal "135" of pixel b. As a result, the screen b is divided into luminance signal values of "142". Similarly, the residual luminance signal of pixel b=“14” is added to pixel c, and the luminance signal after the addition process becomes “149”, and its representative value is encoded as “144”. As described above, by adding the residual luminance signal, which is the difference between the luminance signal value and the representative value, to the luminance signal value of adjacent pixels, the average luminance signal is It is encoded as an image with a value of “132”. That is, in the above description, the brightness signal values of all four pixels are represented by "128", but this example differs in that the brightness of pixel c is represented by "144". According to human vision, the brightness of adjacent pixels is perceived as being averaged, so by applying the processing of this application that adds the differences,
It is possible to create an image that is closer to the original image without any deterioration. Reference numeral 1 denotes a scanning reading device, and a quantizing device 2 quantizes the analog luminance signal of the image read by the scanning reading device. The quantizer 2 is a block according to the invention. 3 is a storage device such as a magnetic disk,
The image signal quantized by the quantizer 2 is stored here and read out as needed. Reference numeral 4 denotes a digital-to-analog converter which converts the read quantized image signal into an analog signal, which is printed by a recording device 5 to restore the original image. Note that the storage device 5 can also be replaced with a facsimile transmission line. In this case, the scanning reading device 1 and the quantization device 2 serve as transmitting terminals, and the digital-to-analog converting device 4 and recording device 5 serve as receiving terminals. Next, referring to FIG. 2, a quantization method in an embodiment of the present invention will be described. FIG. 2 is a detailed block diagram of the quantization device 2 shown in FIG. 1. The read luminance signal obtained by scanning is input to an adder circuit 21 via a line 20. The storage circuit 22 stores the amount to be allocated to and added to the pixel out of the residual signal from the pixel processed before the pixel, and when the pixel requires it,
Output that amount on line 23. Therefore, addition circuit 2
1 adds the signals on lines 20 and 23 to create a luminance signal and outputs it on line 24. Judgment circuit 25
determines in which area the luminance signal value is located,
The determination result is output on a line 26 as an area representative value and on a line 29 as a quantized value. The differential circuit 27 connects the line 26
The difference between the upper area representative value and the luminance signal value on the line 24 is determined and sent to the storage circuit 22 via the line 28. Then, as described above, the storage circuit 22 allocates the residual luminance signal of the pixel to one or more pixels that are processed after the pixel. When there is one pixel to be allocated, the pixel to be processed next after the pixel is allocated, and if there are two pixels to be allocated, the pixel to be processed next to the pixel and the pixel at the same position as the pixel after one scanning line. A preferable example is to allocate approximately 1/2 of the residual brightness to each of the following. The quantized value outputted from the determination circuit 25 to the line 29 is sent to the storage device 3 shown in FIG. By sequentially performing the above processing on all pixels, the range that the pixel read luminance signal can take is divided into n areas with unequal intervals, and the previously processed pixel The residual brightness signals of the pixel are added to form a brightness signal, the brightness signal is quantized depending on which area it falls into, and the residual brightness signal, which is the difference between the representative value of the area and the brightness signal, is processed after the pixel. It can be added to the read luminance signal of the pixel to be read. Next, the quantization method in one embodiment of the present invention will be explained in more detail with reference to the table. Tables 1 and 2 explain examples of dividing areas when the number of areas is n=8.

【table】

【table】

[Table] In Table 1, in column A, specific read brightness values are numbers given for convenience and are associated with 15 read brightness values from 0 to 14. Column B is a set of read brightness values when the lowest read brightness excluding the read brightness of 0 is set to 1 and the read brightness difference ratio is set to 1.49. At this time, the maximum read brightness is 173.27. Note that brightness is a physical quantity and usually has units, but
In the description of this embodiment, units will be omitted and quantities will be expressed numerically. Column C reads the value of column B and normalizes the maximum brightness by 1.000, and column D reads the value of column B and calculates the maximum brightness.
This is the case when normalized by 255. When the number n of areas is 8, (2n-1) read brightness values are determined. The values at both ends and every other value between them are the area representative values, and their read luminance numbers are indicated by circles. In Table 2, column E is the quantization value given to each area representative value, and the number of bits in binary representation is m.
When n=8, m=3 bits in binary notation, and the case where the number of areas is 8 is shown.
Column F is the range of read luminance values for each quantized value. In this embodiment, (2n-1) reading brightness values are set in order of value from the reading brightness range that the image signal can take, and from 0 to (2n-2)
Assign numbers up to, number 0 and number (2n-
2) is the area representative value, and the area representative value of number 0 is the area representative value of numbers 1 and below, and the area representative value of number 2 is the area representative value, such that it represents the reading brightness range of numbers 1 to 3. A region relationship is established.
Note that the area representative values do not necessarily have to be taken at both ends.
Although the object of the present invention can be achieved to some extent, the object of the present invention can be completely achieved if taken at both ends.
In other words, the total luminance considered within a certain area range is
This means that it does not change despite the decrease in the number of codes. Note that in Tables 1 and 2, the brightness values and ranges are expressed as analog quantities, but in the present invention, it does not matter whether the areas are drawn in analog quantities or digital quantities. Table 3 shows the relationship between the quantized value (column e) and the read luminance range (column f) when the input read luminance signal is a digital value.

【table】

[Table] Column e is a binary representation, and column f is a decimal representation. The read brightness value corresponding to the read brightness number may be, as described above, when the difference in brightness values is geometric, or it may be a value determined by an appropriate rule or trial as necessary. Furthermore, depending on the value or sign of the residual luminance signal, the composite luminance signal value may exceed the maximum value of the read luminance signal or take a negative value. Effects of the Invention According to the present invention, it is possible to quantize image information such as a photograph with halftones with a small number of codes and with little deterioration. Therefore, when storing an image, the memory capacity is reduced and the data is transmitted. In this case,
Transmission can be carried out in a short period of time, and the effect is great.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a system for implementing a method for quantizing multivalued image information in one embodiment of the present invention, and FIG. 2 is a block diagram of a quantization device. DESCRIPTION OF SYMBOLS 1... Reading device, 2... Quantization device, 3... Storage device, 4... D/A conversion device, 5... Recording device, 21... Addition circuit, 22... Storage circuit, 25
...Judgment circuit, 27...Difference circuit.

Claims (1)

[Claims] 1. Divide the range that the read luminance signal of a pixel can take into n areas with unequal intervals, and add the residual luminance signal of the previously processed pixel to the read luminance signal of the pixel to obtain a luminance signal; The luminance signal is quantized depending on whether it falls into any of the regions, and a residual luminance signal, which is the difference between the representative value of the region and the luminance signal, is added to the read luminance signal of a pixel processed after the pixel. Features a quantization method for multivalued image information.