WO2006105721A1 - Procede et appareil capable de production de points grilles de modulation de frequence a grande vitesse - Google Patents

Procede et appareil capable de production de points grilles de modulation de frequence a grande vitesse Download PDF

Info

Publication number
WO2006105721A1
WO2006105721A1 PCT/CN2006/000567 CN2006000567W WO2006105721A1 WO 2006105721 A1 WO2006105721 A1 WO 2006105721A1 CN 2006000567 W CN2006000567 W CN 2006000567W WO 2006105721 A1 WO2006105721 A1 WO 2006105721A1
Authority
WO
WIPO (PCT)
Prior art keywords
error
value
buffer register
pixel
line
Prior art date
Application number
PCT/CN2006/000567
Other languages
English (en)
French (fr)
Inventor
Zhihong Liu
Feng Chen
Bin Yang
Original Assignee
Peking University Founder Group Co., Ltd
Beijing Founder Electronics Co., Ltd.
Peking University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Peking University Founder Group Co., Ltd, Beijing Founder Electronics Co., Ltd., Peking University filed Critical Peking University Founder Group Co., Ltd
Priority to EP06722219.0A priority Critical patent/EP1874031B1/en
Priority to JP2008504606A priority patent/JP4481343B2/ja
Priority to US11/918,077 priority patent/US8363278B2/en
Publication of WO2006105721A1 publication Critical patent/WO2006105721A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04NPICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N1/00Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
    • H04N1/40Picture signal circuits
    • H04N1/405Halftoning, i.e. converting the picture signal of a continuous-tone original into a corresponding signal showing only two levels
    • H04N1/4051Halftoning, i.e. converting the picture signal of a continuous-tone original into a corresponding signal showing only two levels producing a dispersed dots halftone pattern, the dots having substantially the same size
    • H04N1/4052Halftoning, i.e. converting the picture signal of a continuous-tone original into a corresponding signal showing only two levels producing a dispersed dots halftone pattern, the dots having substantially the same size by error diffusion, i.e. transferring the binarising error to neighbouring dot decisions

Definitions

  • the present invention relates to a method and apparatus for generating a halftone dot in a digital image halftone field, and more particularly to a method and apparatus for generating a frequency modulation network point at a high speed.
  • the digital image halftone technology is divided into two types, namely, amplitude modulation hanging net and frequency modulation hanging net.
  • the error diffusion method is the most commonly used method, and the error diffusion method is to put the original image every
  • the gray value of the image points is compared with the threshold to produce a halftone dot, and the error between the gray value of the pixel and the threshold is diffused to the image point around the pixel, for example, for a 256-level gray scale
  • the image has a threshold of 127 and a pixel with a gray value of 150.
  • the image should be recorded as white, but in fact the image is not really white, and there is gray between white and white.
  • the difference is 23, and the error of 23 is dispersed in a certain way to the image points around the pixel.
  • the algorithm first compares the gray value of the current pixel with the threshold, records the pixel as 1 or 0, ie white or black, and then compares it with the interpretation value, calculates the error, assigns the error to the surrounding image points, and modifies the surrounding image.
  • the gray value of the point specifically, the Floyd-Steinberg algorithm adds 7/16 of the error to the first image point on the right side of the current image point, and the error 3-6 is added to the first image point on the left side of the next line.
  • the error of the 8 8 is added to the image point of the next line, and the error of 1 8.6 is added to the first image point on the right side of the next line, so that the error of the current image point is dispersed to the surrounding image points. This process is repeated to perform such halftone and gray value correction for each image point in the image.
  • the method of error diffusion usually produces better results, but its shortcomings are obvious. This method requires a lot of calculations and more memory operations, resulting in slower dot gain, Stucki algorithm and S-scan.
  • the specific implementation usually uses a memory to save the result of the error diffusion of the current processing line and the following two lines. Generally speaking, it is an error line, and each position of the error line represents the accumulated error value at the corresponding position, so that For each pixel processed, it is necessary to read the original error cumulative value corresponding to the current pixel of the error row and the original error cumulative value at the adjacent 12 positions, and calculate the newly generated error value of the current pixel, and then assign the value.
  • an object of the present invention is to provide a method and apparatus capable of generating a frequency modulation network point based on an error diffusion method at a high speed, and the apparatus using the method only needs to perform one time for the error line memory.
  • the read and write operations greatly increase the speed of the FM network.
  • the technical solution adopted by the present invention is: A method capable of generating a frequency modulation network point at a high speed, and the method specifically includes the following steps:
  • the number of lines of the error line is 2 nth power, and the bus width of the error line memory can ensure that n times of cells can be read and written from the error line memory at a time, and the number of lines of the error line is greater than or equal to one pixel.
  • the number of lines to which the error can be spread, the error line is repeatedly used, and the above n is a positive integer;
  • the error line content is initialized to zero. Further, in order to make the invention have better invention effect, the following method is adopted when setting the error buffer register stack in step b, including the following steps: 1)
  • the error buffer register file includes two parts: an error allocation buffer register file and an error accumulation buffer register file, wherein the error allocation buffer register heap size can save all error distribution values of all relevant pixels at a certain position of the error line, where The associated pixel is the current processing intra-row error value that can be diffused to all pixels of the error row position.
  • the error accumulation buffer register heap size is sufficient to hold all relevant error accumulation values in the original error row of the lower row of the currently processed read line.
  • the correlation error accumulation value is the original error accumulation value in all the error lines read out from the original error accumulation value at a certain position of the read error line to the new error accumulation value at the position;
  • step d Initialize the error buffer register file, ie the error buffer register heap content is initialized to 0 at the beginning of each line of processing. Furthermore, in order to make the present invention have a better invention effect, the following method is used in step d to generate a network point according to the current source pixel and the error line and the contents of the error buffer register file, including the following steps:
  • step e Adding the original error cumulative value at the corresponding position of the current pixel to S, and adding the final pixel value to the current source pixel value, and comparing the value with the threshold to generate a final halftone point. Furthermore, in order to make the invention have a better invention effect, the following method is used in the step e to update the error line and the error buffer register stack, including the following steps:
  • the present invention also proposes a device for implementing the method using a hardware circuit, the device 1 comprising an error line memory 2, an error buffer register file 3, a threshold comparison circuit 4, an error generation circuit 5, and an error buffer
  • the register file control circuit 6, the gray level generating circuit 7, and the error line control circuit 8 are seven; further, the error line memory 2 is used to store the error diffusion result during processing; the error buffer register file 3 is used to buffer the intermediate error value.
  • the gradation generation circuit 7 is connected to the source pixel, the error line memory 2, and the error buffer register file 3 for generating the final gradation after the error diffusion of the source pixel; the comparison value comparison circuit 4 is connected to the gradation generation circuit 7
  • the threshold and final grayscale comparison produces a final halftone;
  • the error generation circuit 5 is coupled to 'the grayscale generation circuit 7 calculates the current error distribution value from the threshold and the final grayscale;
  • the error buffer register file control circuit 6 controls the error buffer register file updates and outputs its input from the content error generation circuit 5 and the error row control circuit 8 and the error buffer register file 3
  • the error line control circuit 8 is connected to the error line memory 2 for generating the read/write address, the read/write control signal and the read/write data of the error line memory, and the input thereof is from the coordinates including the currently processed pixel and the error buffer register file.
  • the gradation generation circuit 7 derives the final pixel gradation based on the contents of the current processing pixel and the error line memory 1 and the error buffer register file 3, and then passes through the threshold comparison circuit. 4 generating a dot, and the final pixel gray scale is further obtained by the error generating circuit 5 to obtain an error distribution value corresponding to the pixel, and then the error buffer register file is completed by the error buffer register file control circuit 6 and the error row control circuit 8, respectively. 3 and one update of the error line memory 2.
  • the error buffer register stack in the above device is implemented by a register file composed of flip-flops, which comprises an error distribution buffer register file and an error accumulation buffer register stack, wherein
  • the error allocation buffer register heap size is sufficient to store all error distribution values of all relevant pixels at a certain position of the error line, wherein the current correlation pixel is the current processing intra-row error value can be diffused to all pixels of the error line position; error accumulation buffer register The heap size is sufficient to hold all associated error accumulated values in the original error line of the lower row of the current processed line read out, wherein the associated error cumulative value is a new error from the original error accumulated value at a certain position of the read error line to the position The accumulated value is written to the original error accumulated value in all the read error lines in the process.
  • the gray scale generating circuit in the above apparatus is composed of an adder whose input is from the source pixel, the current error value read in the error line and the error buffer register file.
  • Current error buffer value it will offset the current source pixel value and the error value corresponding to the position read in the error line and the error of the previous pixel of the current pixel buffered in the error buffer register stack to the current pixel
  • the values are added to give the final current pixel gray value.
  • the error buffer register stack control circuit in the above device mainly consists of n multiplexer groups, where n is the number of registers in the error buffer register file.
  • the input of the multiplexer includes the current error distribution value.
  • the output of the multiplexer is the updated value of the register in the error buffer register file.
  • the error line control circuit in the above device is composed of a read address register, a write address register, a read/write control circuit and a write data generation circuit, wherein the read address register and the write address register are composed of The column number of the current pixel is generated, and the write data generating circuit is composed of an adder, wherein the input of the adder is derived from the original error accumulated value, the current error assigned value, and the associated error buffer value.
  • the effect of the present invention is: In the field of digital image halftone, when the frequency modulation dot is generated by the error diffusion based method, the method and the device only need to perform one memory read and one memory write operation for processing one pixel, and can use dedicated The hardware circuit is realized, which greatly improves the generation speed of the frequency modulation network.
  • Figure 1 is a schematic diagram of the error diffusion principle based on the Floyd-Steinberg algorithm
  • Figure 2 is a schematic diagram of error expansion based on the Stucki algorithm
  • Figure 3 is an illustration of an error line memory with an image width of 7 pixels, usually based on the Stucki algorithm
  • FIG. 5 is a schematic diagram of an error line memory with an image width of 7 pixels based on the Stucki algorithm in the method of the present invention
  • FIG. 6 is a schematic diagram of an error buffer register stack based on the Stucki algorithm in the method of the present invention
  • FIG. 7 is a block diagram of an apparatus for generating a frequency modulation network point at a high speed according to the present invention
  • FIG. 8 is a block diagram of a gradation generation circuit in the apparatus for generating a frequency modulation network point at a high speed according to the present invention
  • FIG. 9 is a block diagram of an error line control circuit in the apparatus for generating a frequency modulation network point at a high speed according to the present invention
  • a block diagram of a multiplexer in an error buffer register file control circuit in the apparatus for generating a frequency modulation network point at a high speed
  • Figure 11 is a schematic diagram of the read and write process of the error line memory when the method of the present invention processes a pixel. detailed description
  • 1 is a schematic diagram of the error diffusion principle based on the Floyd-Steinberg algorithm.
  • the error of the current pixel* is assigned to the four adjacent pixels shown in FIG. 1 in the scale shown in FIG. 1.
  • 2 is a schematic diagram of the error diffusion principle based on the Stucki algorithm.
  • the error of the current pixel* is assigned to the 12 adjacent pixels shown in FIG. 1.
  • the error distribution ratio of some positions at the 12 positions is the same.
  • FIG. 3 is a schematic diagram of an error line memory with an image width of 7 pixels, which is usually based on the Stucki algorithm, wherein (i, j) represents the error line position corresponding to the pixel of the i-th row and the j-th column, which is easy to know based on Stuck i
  • the error line of the algorithm can be recycled, and only 3 lines of error lines are needed, wherein the pixels of the i + 3rd line and the i-th line use the same error line.
  • a specific implementation method for processing an image by using a method capable of generating a frequency modulation network point at a high speed includes the following steps:
  • the figure is a schematic diagram of an error line memory based on the Stucki algorithm when the error line memory is organized and initialized in the step (1) of the above method, wherein the image width is 7 pixels, and (i, j) indicates i line, the error line position corresponding to the pixel of the jth column.
  • the data bus width of the memory is often an integer power of 16, 32 or 64, etc.
  • the number of lines of the error line used in the present invention is also defined as an integer power of 2, specifically to the Stucki algorithm-based method, the error behavior in this embodiment is 4 lines, which is used cyclically, that is, the pixels of the i + 4th row and the i-th row use the same error row.
  • the method of the present invention is different from the prior art in that, in the method of the present invention, the error lines are arranged in columns, that is, the column numbers of the same four columns of error row units are adjacently arranged. In this way, wrong The difference line memory uses a 32-bit bus width to read and write all the accumulated error values at the same column position in the error line. As shown in FIG.
  • the figure is a schematic diagram of an error buffer register file based on the Stucki algorithm when the error buffer register is initialized in the above method step (2), which includes several previous ones in the same row of the currently processed pixel and the currently processed pixel.
  • the values of all the error distributions of the associated 4 pixels and the values of the number of original error lines of the lower row of the current processed line read are taken as an example of the Stucki algorithm shown in FIG. 2 because the error of one position of the error line may be Up to 5 pixels from the current line, so the error buffer register heap needs to store all the error distribution values of the 5 adjacent pixels in the same row, and because the error of the currently processed pixel can only go to the top of the next row and the first two rows.
  • the pixels are diffused, and it is easy to know that only the final error value of the next line of the current line is related to the original value in the error line memory, and the error value of the lower two lines of the current line is 0 before the current line starts processing, and the current line
  • the writing of the final error cumulative value at a certain position in the error line of the next line is performed after the error line corresponding to the next 2 pixels is read, which is performed.
  • the accumulated error value of the read error line is three, so it is only necessary to save the value of the original error line of the next line of the three current lines in the error buffer register file. As shown in FIG.
  • the following method is used to generate a network point according to the current source pixel and the error line and the contents of the error buffer register file, including the following steps:
  • the method for updating the error line and the error buffer register file in the step (5) of the method of the present invention adopts the following method, including the following steps:
  • the error generated by the current pixel is assigned to the value Dj of all adjacent pixels according to the final pixel value and the threshold value.
  • the Stucki algorithm shown in 2 is taken as an example.
  • the current processing pixel is P5
  • PL2 PiDi + P2D2 + P3D3 + P4D2 + P5Dl.
  • a device 1 capable of generating a frequency modulation network at a high speed as shown in FIG. 7, the device includes an error line memory 2, an error buffer register file 3, an analysis value comparison circuit 4, an error generation circuit 5, and an error buffer register file control circuit. 6.
  • the gray level generating circuit 7 and the error line control circuit 8 have a total of seven parts.
  • the gradation generation circuit 7 derives the final pixel gradation based on the contents of the current processing pixel and the error line memory 2 and the error buffer register file 3, and then generates a dot by the threshold comparison circuit 4, and the final pixel ash
  • the error distribution circuit 5 also obtains an error distribution value corresponding to the pixel, and then the error buffer register stack circuit 3 and the error line control circuit 8 respectively perform an update of the error buffer register file 3 and the error line memory 2, respectively.
  • the error line memory 2 is a column-organized memory for storing an error accumulation value which is connected to the error line control circuit 8 and controlled by the error line control circuit 8;
  • the error buffer register file 3 is composed of a hardware flip-flop for buffering the intermediate error distribution value, which is connected to the error buffer register file control circuit 6 and controlled by the error buffer register file control circuit 6;
  • the grayscale generating circuit 7 is used to generate the final gray level of the source pixel, its input is connected to the source pixel and the error buffer register file control circuit 6, and its output is supplied to the value comparison circuit 4 and the error generating circuit 5;
  • 8 is a block diagram of a grayscale generating circuit in a device capable of generating a frequency modulation dot at a high speed according to the present invention, which generates an imaginary value of a final pixel by an adder, wherein a gray value of the final pixel is equal to a value of the source pixel.
  • the threshold comparison circuit 4 is configured to compare the gradation generated by the gradation generation circuit ⁇ with a threshold to generate a dot, and its input is connected to the gradation generation circuit 7, and the output is the final dot;
  • the error generating circuit 5 is for generating the error distribution value of the current pixel by calculating or looking up the gradation generated by the gradation generating circuit 7, its input is connected to the gradation generating circuit 7, and the output is supplied to the error buffer register.
  • the error buffer register file control circuit 6 is used to control the update of the error buffer register file, and its input is connected to the error generation circuit 5 and the error line control circuit 8 and the error buffer register file 3, and its output is connected to the error line control.
  • FIG. 10 is a block diagram of the multiplexer in the error buffer register file control circuit of the device capable of generating the FM network point at high speed according to the present invention, the error
  • the buffer register file control circuit uses n multiplexers (mul t iplexer ) to complete an update of the error buffer register file and the initial clear at the beginning of each line in each pixel processing cycle, where n is the error buffer.
  • the number of registers in the register file that is, each register in the corresponding error buffer register stack contains a multiplexer of the same structure, wherein the input of the multiplexer includes all error distribution values of the currently processed pixels, that is, the current error distribution. Value, read the current error accumulation value of the next row corresponding to the current processing pixel from the error line and the error buffer register file.
  • the output of the multiplexer can be represented as corresponding to PiDj
  • the output of the multiplexer is Pi + iDj
  • the multiplexer output is Li.
  • the error line control circuit 8 is used to control the reading and writing of the error line memory 1. Its input is from the error generating circuit 5 and the error buffer register file control circuit 6, and its output is connected to the gray level generating circuit 7, and it also has an error. Line memory 2 is connected.
  • 9 is a block diagram of an error line control circuit in a device capable of generating a frequency modulation network point at a high speed according to the present invention, which is composed of a read address register, a write address register, a read/write control circuit, and a write data generation circuit, wherein The read address register, the write address register is generated by the current processing pixel coordinates, and the write data generation circuit is composed of an adder, which is written by the original error cumulative difference, the current error distribution value and the associated error buffer.
  • Figure 11 is a schematic diagram of the read and write process of the error line memory when the method of the present invention processes a pixel.
  • the current processing pixel is shown as *, as in the prior art, when processing the current processing pixel *, it is necessary to read PL2 and PLi, Lo in the error line shown in FIG. And the original content of Li and all the positions marked as X, after the current pixel is processed, it is also necessary to update the accumulated values at the 12 positions adjacent to the current pixel, and the method and apparatus according to the present invention are used to process the current pixel.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Facsimile Image Signal Circuits (AREA)
  • Image Processing (AREA)

Description

一种能够高速产生调频网点的方法和装置
技术领域
本发明涉及一种数字图像半色调领域的网点生成方法和装置, 具体涉及一种 能够高速产生调频网点的方法和装置。 背景技术
数字图像半色调技术分为两种, 分别是调幅挂网和调频挂网, 而在通常的调. 频挂网方法中, 误差扩散方法是最常用的方法, 误差扩散方法是把原图像的每个 像点的灰度值与阔值相比较产生半色调点的同时,把像点的灰度值与阈值之间的误 差扩散到该像点周围的像点上,例如对于一个 256级灰度的图像,阈值为 127,有一个 像点的灰度值为 150,经比较可知,该像点应记为白色,但实际上该像点并不是真正 的白色,与白色之间存在的灰度差为 23,将 23这个误差按一定的方法分散到该像点 的周围的像点上。
'误差扩散到周围点的方法有多种,其中最著名及常见的是 Floyd- Steinberg算 法。 该算法首先比较当前像素点的灰度值与阈值,将该像点记为 1或 0,即白色或黑 色,然后与阐值进行比较, 计算误差,分配误差到周围的像点,修改周围像点的灰度 值, 具体而言, Floyd- Steinberg算法中是将误差的 7/16加到当前像点右边第一 个像点上,误差的 3八 6加到下一行左边第一个像点上,误差的 5八 6加到下一行正对 的像点上,误差的 1八 6加到下一行右边第一个像点上,这样把当前像点的误差分散 到周围的像点上, 反复进行该过程,对图像中的每个像点进行如此的半色调化和灰 度值的修正即可。
在 Floyd- Steinberg算法的基础上, 又出现了一些改进的算法, 如 Stucki算 法, 该算法将当前像素产生的误差扩散到当前像素点的相邻 12个点上, 由于涉及 了更多的点,输出图像效果好,同时,在误差扩散时,若总是从左到右一个扫描行一 个扫描行的处理每个像点,则会形成一个扫描行的误差简单的加至下一行,从而引 起误差的堆积,表现在输出图形上是图像有一种驱赶的趋势, 所以在扫描时,往往 采用 S形的扫描方式,即如果是奇数行则从左到右扫描, 如果是偶数行则从右到左 扫描,依次类推,直至最后一行, 这样能取得更好的效果。
采用误差扩散的方法通常能产生较好的效果, 但它的不足也是显而易见的, 即这种方法需要大量的计算及比较多的存储器操作, 产生网点的速度慢, 以 Stucki算法和采用 S型扫描方式为例, 具体实现时通常是采用一个存储器来保存 当前处理行及下面两行的误差扩散的结果, 一般称它为误差行, 误差行的每一个 位置代表对应位置上的误差累积值, 这样每处理一位像素, 需要读出误差行的当 前像素所对应的原有误差累积值及相邻 12个位置上的原误差累积值, 计算出当前 像素新产生的误差值后, 将该值分配到相邻 12个位置上并且与该位置上原误差累 积值相加, 并将结果写入误差行的这 12个位置上。 由此可见, 现有的方法需要大 量的存储器操作, 而存储器的读写速度在计算机系统中往往是系统速度的瓶颈, 因此虽然基于误差扩散的挂网方法能取得较好的效果, 却因为速度的原因, 应用 范围受到限制, 在通常的打印机, 复印机等设备中, 虽然通常用专用硬件来提高 基于误差扩散方法的网点产生速度, 但因为现有技术含有大量存储器操作, 速度 仍然受到影响,'难以达到高速打印机和复印机的实时性要求, 使基于误差扩散的 挂网方法难以在现代的高速打印机和复印机中得到充分应用。 发明内容
针对现有技术中存在的缺陷, 本发明的目的是提供一种能够高速产生基于 误差扩散方法的调频网点的方法和装置, 采用该方法的装置每处理一位像素对 误差行存储器只需要进行一次读和一次写的操作, 大大提高调频网点的产生速 度。 为达到以上目的, 本发明采用的技术方案是: 一种能够高速产生调频网点的 方法, 该方法具体包括以下步骤:
a )按列组织并初始化误差行存储器;
b )设定误差緩冲寄存器堆;
c )获取源像素;
d )依据当前源像素及误差行和误差緩冲寄存器堆的内容产生网点; e )更新误差行及误差緩冲寄存器堆;
f) 重复步驟 c , d, e直到一行结束;
g) 重复步驟 b, c , d, e, f 直到图像结束。 更进一步, 为使本发明具有更好的发明效果, 步蘇 a中按列组织并初始化误 差行存储器时采用如下方法, 包括如下步驟:
1 )误差行的行数为 2的 n次方, 并且误差行存储器的总线宽度能够保证一次 可以从误差行存储器中读写 1 的 n次方个单元, 同时误差行的行数大于等于一个 像素的误差所能够扩散到的行数, 误差行循环重复使用, 上述的 n为正整数;
2 ) 列号相同的每 的 n次方列误差位置相邻排列;
3 )误差行内容被初始化为 0。 更进一步, 为使本发明具有更好的发明效果, 步驟 b中设定误差緩冲寄存器 堆时采用如下方法, 包括如下步骤: 1 )误差緩冲寄存器堆包括误差分配緩冲寄存器堆和误差累积緩冲寄存器堆两 部分, 其中误差分配緩冲寄存器堆大小能够保存误差行某一位置的所有相关像素 所有误差分配值, 其中当前相关像素为当前处理行内误差值能扩散到这一误差行 位置所有像素, 误差累积緩冲寄存器堆大小为足够保存读出的当前处理行的下面 行的原误差行中的所有相关误差累积值, 其中相关误差累积值为从读出误差行某 一位置上原误差累积值到该位置新的误差累积值被写入过程中所有读出的误差行 中的原误差累积值;
2 )初始化误差緩冲寄存器堆即每行处理开始时误差緩冲寄存器堆内容被初始 化为 0。 更进一步, 为使本发明具有更好的发明效果, 步驟 d中依据当前源像素及误 差行和误差緩冲寄存器堆的内容产生网点时采用如下方法, 包括如下步骤:
1 )从误差行存储器中一次读出对应当前源像素位置及下面行同一列位置上的 误差累积值;
2 ) 由误差緩冲寄存器堆的内容计算出当前源像素的同一行的所有以前的像素 处理所产生的误差分配到当前源像素的误差值的和 S;
3 )将当前像素对应位置上的原误差累积值加上 S, 再和当前源像素值相加得 出最终像素值, 将该值和阈值比较产生最终网点。 更进一步, 为使本发明具有更好的发明效果, 步骤 e中更新误差行及误差緩 冲寄存器堆时采用如下方法, 包括如下步驟:
1 )根据最终像素值和阈值得出当前像素产生的误差分配到所有相邻像素的值
Dj ;
2) 误差分配緩冲寄存器堆中每一像素的误差分配值更新为后一像素的对应 值, 而最后一个像素的误差分配值更新为 Dj ;
3 )误差累积緩冲寄存器堆中每一像素的误差累积值更新为后一像素的对应 值, 而最后一个像素的误差累积值更新为当前像素对应位置下面行上的原误差累 积值;
4 )根据误差緩冲寄存器堆内容得出误差行的一列的所有写入数据值并一次写 入误差行存储器中, 其中误差行一列中某一点的写入数据为存放在误差緩冲寄存 器堆中的扩散到该点的所有误差分配值和该点原误差累积值之和。 为实现上述方法, 本发明还提出了一个采用硬件电路来实现该方法的装置, 该装置 1包括误差行存储器 2 , 误差緩冲寄存器堆 3, 阈值比较电路 4, 误差生成电 路 5 , 误差緩冲寄存器堆控制电路 6 , 灰度生成电路 7 , 误差行控制电路 8七部分; 进一步, 误差行存储器 2用来保存处理过程中的误差扩散结果; 误差緩冲寄存 器堆 3用来緩冲中间误差值; 灰度生成电路 7连接到源像素、 误差行存储器 2和误差 緩冲寄存器堆 3 , 用来产生源像素经过误差扩散后的最终灰度; 阐值比较电路 4连 接到灰度生成电路 7将阈值和最终灰度比较产生最终网点; 误差生成电路 5连接到' 灰度生成电路 7由阈值和最终灰度计算出当前误差分配值; 误差緩冲寄存器堆控制 电路 6来控制误差緩冲寄存器堆的更新及输出, 它的输入来自误差生成电路 5和误 差行控制电路 8和误差緩冲寄存器堆 3的内容反馈; 误差行控制电路 8连接到误差行 存储器 2用来产生误差行存储器的读写地址及读写控制信号及读写数据, 它的输入 来自包括当前处理象素的坐标及误差緩冲寄存器堆控制电路 6及误差生成电路 5; 该装置工作时, 灰度生成电路 7根据当前处理像素和误差行存储器 1及误差 緩冲寄存器堆 3的内容得出最终的像素灰度, 然后通过阈值比较电路 4产生网 点, 同时最终的像素灰度还通过误差生成电路 5得出对应该像素的误差分配值, 然后通过误差緩冲寄存器堆控制电路 6和误差行控制电路 8来分别完成误差緩冲 寄存器堆 3和误差行存储器 2的一次更新。 更进一步, 为使本发明具有更好的效果, 上述装置中误差緩沖寄存器堆釆用 采用由触发器组成的寄存器堆实现, 它包括误差分配缓冲寄存器堆和误差累积緩 冲寄存器堆两部分, 其中误差分配緩冲寄存器堆大小为足够保存误差行某一位置 的所有相关像素所有误差分配值, 其中当前相关像素为当前处理行内误差值能扩 散到这一误差行位置所有像素; 误差累积緩冲寄存器堆大小为足够保存读出的当 前处理行的下面行的原误差行中的所有相关误差累积值, 其中相关误差累积值为 从读出误差行某一位置上原误差累积值到该位置新的误差累积值被写入过程中所 有读出的误差行中的原误差累积值。 更进一步, 为使本发明具有更好的效果, 上述装置中灰度生成电路由一个加 法器组成, 加法器的输入来自源像素, 误差行中读出的当前误差值和误差緩冲寄 存器堆中的当前误差緩冲值; 它将当前源像素的值及误差行中读出的对应该位置 的误差值及误差緩冲寄存器堆中緩冲的当前像素同一行的以前像素对当前像素的 误差緩冲值相加而得出最终的当前像素灰度值。 更进一步, 为使本发明具有更好的效果, 上述装置中误差緩冲寄存器堆控 制电路主要由 n 个多路器组 , 其中 n为误差緩冲寄存器堆中的寄存器个数, 它用来产生误差缓冲寄存器堆的输入及控制信号从而完成在每一像素处理周期 内误差緩冲寄存器堆的一次更新及每一行开始时的初始化清零, 多路器的输入 包括当前误差分配值, 当前误差累积值及误差緩冲寄存器堆中的原值, 多路器 的输出为误差緩冲寄存器堆中寄存器的更新值。 更进一步, 为使本发明具有更好的效果, 上述装置中误差行控制电路由读 地址寄存器, 写地址寄存器, 读写控制电路及写入数据生成电路组成, 其中读 地址寄存器和写地址寄存器由当前像素的列号产生, 写入数据生成电路则由加 法器组成, 其中加法器的输入来自于原误差累积值, 当前误差分配值及相关误 差緩冲值。 本发明的效果在于: 在数字图像半色调领域, 利用基于误差扩散的方法产 生调频网点时, 本方法和装置每处理一个像素只需进行一次存储器的读和一次 存储器的写操作, 并且能够采用专用硬件电路来实现, 大大提高了调频网点的 产生速度。 附图说明
图 1是基于 Floyd- Steinberg算法的误差扩散原理示意图;
图 2 是一个基于 Stucki算法的误差扩 «理示意图;
图 3 是通常基于 Stucki算法的一个图像宽度为 7个像素的误差行存储器示 意图;
图 4 是采用本发明所述方法处理一个图像的流程框图;
图 5 是本发明所述方法中基于 Stucki算法的一个图像宽度为 7个像素的误 差行存储器示意图;
图 6 是本发明所述方法中基于 Stucki算法的误差緩冲寄存器堆示意图; 图 7是本发明所述的高速产生调频网点的装置的框图;
图 8是本发明所述的高速产生调频网点的装置中灰度生成电路的框图; 图 9是本发明所述的高速产生调频网点的装置中误差行控制电路的框图;, 图 10是本发明所述的高速产生调频网点的装置中误差緩冲寄存器堆控制电 路中多路器的框图;
图 11 是本发明所述的方法在处理一个像素时误差行存储器的读写过程示意 图。 具体实施方式
下面结合说明书附图和具体实施方式对本发明作进一步地描述。 附图 1是基于 Floyd- Steinberg算法的误差扩散原理示意图, 当前像素 *的 误差按附图 1所示的比例被分配到附图 1所示的 4个相邻像素上去。 附图 2是基于 Stucki算法的误差扩散原理示意图, 当前像素 * 的误差被分配 到附图 1所示的 12个相邻像素上去,通常这 12个位置上的某些位置的误差分配比 例是一样的, 附图 2所示的算法只需要 5个不同的误差分配比例, 其中可取 Dl = 1/44, D2 = 2/44, D3 = 5/44, D4 = 4/44, D5 = 8/44 的比例来分配当前像素 *的误 差。 通常采用这种算法并且采用 S 型的扫描处理时需要有误差行来保存当前处理 像素对下面 2 行的误差分配累加值, 这样误差行通常为 3 行, 宽度为源图像宽 度。 容易得知, 通常处理一个像素需要大量的误差行读写操作来更新误差行存储 器中 12个位置上的误差累积值。 附图 3是通常基于 Stucki算法的一个图像宽度为 7个像素的误差行存储器 示意图, 其中 (i , j )表示第 i行, 第 j列的像素对应的误差行位置, 容易得知 基于 Stuck i算法的误差行可以循环使用, 只需要 3行误差行即可, 其中第 i + 3 行和第 i行的像素使用同一误差行。 如图 4所示, 本发明所述一种能够高速产生调频网点的方法用于处理一个图像的 具体实现方法, 包括如下步骤:
( 1 )按列组织并初始化误差行存储器;
( 2 )设定误差緩冲寄存器堆;
( 3 )获取源像素;
( )依据当前源像素及误差行和误差緩冲寄存器堆的内容产生网点;
( 5 )更新误差行及误差緩冲寄存器堆;
( 6 )判断是否一行处理结束, 如否, 则转入步驟 3 , 如是则转入步驟 7; ( 7 )判断是否图像处理结束, 如否, 则转入步驟 2 , 如是处理结束。 如图 5所示, 该图是上述方法步驟(1 ) 中按列组织并初始化误差行存储器 时基于 Stucki算法的一个误差行存储器示意图, 其中图像宽度为 7个像素, ( i , j )表示第 i行, 第 j列的像素对应的误差行位置, 在通常的计算机系统中 存储器的数据总线宽度往往是 16 , 32或 64等 2的整数次方, 因此, 本发明采用 的误差行的行数同样定义为 2的整数次方, 具体到基于 Stucki算法的方法,本实 施例中误差行为 4行, 循环使用, 即其中第 i + 4行和第 i行的像素使用同一误 差行。 同时, 如附图 5所示, 本发明所述方法和现有技术不同的是, 本发明所述 方法中, 误差行是按列排列的, 即列号相同 4列误差行单元相邻排列, 这样, 误 差行存储器采用 32位的总线宽度, 就一次可以读写误差行中同一列位置上的所 有误差累积值。 如图 6所示, 该图是上述方法步骤( 2 ) 中初始化误差緩冲寄存器堆时基于 Stucki算法的误差緩沖寄存器堆示意图, 它包含有当前处理像素和当前处理像 素的同一行中之前若干个相关 4象素的所有误差分配值及读出的当前处理行的下面 行的若干个原误差行的值, 以附图 2所示的 Stucki算法为例, 因为误差行的一 个位置上的误差可能来自当前行的最多 5个像素, 所以, 误差缓冲寄存器堆需要 保存同一行 5个相邻像素的所有误差分配值, 同时因为当前处理像素的误差最多 只能往下一行及下两行的前 2个像素扩散, 并且容易得知只有当前行的下一行的 最终误差值与误差行存储器中的原值有关, 当前行的下面笫二行的误差值在当前 行开始处理前为 0, 而当前行的下一行的误差行中某一位置的最终误差累积值的 写入是在它后 2个像素对应的误差行读出之后才进行的, 这个过程中, 读出的误 差行中原误差累积值为 3个, 所以误差緩冲寄存器堆中只需保存 3个当前行的下 一行的原误差行的值即可。 如图 6所示, 本实施例中, 误差緩冲寄存器堆为能够 緩冲相邻 5个像素的所有误差分配值及当前处理行的下一行的 3个原误差行值的 寄存器堆, 其中设 P5为当前处理像素, PiDj ( i, j=1...5 )表示 P5及它的 4个 同一行前面相邻像素的所有误差值, PiLi ( i = 3, 4 , 5 )则 ^示读出的 Pi位置上 下一行的误差行存储器中的原误差累积值。 本发明所述的方法中步骤(4 ) 中依据当前源像素及误差行和误差緩冲寄存器 堆的内容产生网点时采用如下方法, 包括如下步骤:
1 )从误差行存储器中一次读出对应当前源像素位置及下面行同一列位置上的 误差累积值, 因为误差行是按列组织的并且误差行的行数与误差行存储器的总线 宽度存在匹配关系, 因此读地址寄存器和写地址寄存器由当前处理像素的列号得 出, 以实现附图 2所示的 Stucki算法为例, 如当前处理像素列号为 j , 则读地址 寄存器 = 4*j , 因为误差行是以 4 行为周期循环使用的, 所以当读出误差行内容 后, 根据当前处理像素的行号的最低 2 位则可得出当前处理像素所对应的误差行 的原误差累积值, 设为 Lo, 当前处理像素下方一行所对应的误差行的原误差累积 值, 设为 Li;
2 ) 由误差緩冲寄存器堆的内容计算出当前源像素的同一行的所有以前的 像素处理所产生的误差分配到当前源像素的误差值的和 S, 以附图 2 所示的 Stucki算法为例, S等于 P5D5+P4D3 ; 3 )将当前像素对应位置上的原误差累积值 Lo加上 S , 再和当前源像素值相 加得出最终像素值, 将该值和阈值比较即可产生最终网点。 本发明所述的方法中步骤(5 ) 中更新误差行及误差緩冲寄存器堆时采用如下 方法, 包括如下步驟:
1 )根据最终像素值和阈值得出当前像素产生的误差分配到所有相邻像素的值 Dj , 以附图 2 所示的 Stucki 算法为例, 分配比例为 Dl = l/44, D2 = 2/44, D3 = 5/44, D4 = 4/44, D5 = 8/44;
2) 误差分配緩冲寄存器堆中每一像素的误差分配值更新为后一像素的对应 值, 而最后一个像素的误差分配值更新为 Dj , 以附图 1 所示的 Stucki 算法为 例, 即 PiDj赋值为 Pi + iDj , PsDj赋值为 Dj , ( i = 1. . 4, j=l. . 5 ) ;
3 )误差累积緩冲寄存器堆中每一像素的误差累积值更新为后一像素的对应 值, 而最后一个像素的误差累积值更新为当前像素对应位置下面行上的原误差累 积值, 以附图 2所示的 Stuck i算法为例, 即 PiLi赋值为 Pi + iLi , (i=3, 4) , PsLi 赋值为 Li;
4 )根据误差緩冲寄存器堆内容得出误差行的一列的所有写入数据值并一次 写入误差行存储器中, 其中误差行一列中某一点的写入数据为存放在误差緩冲寄 存器堆中的扩散到该点的所有误差分配值和该点原误差累积值之和, 以实现附图
2所示的 Stucki算法为例, 当前处理像素为 P5时, 则可得出在 P5前第 2个像 素位置的所在列的最终写入数据 PLi和 PL2 , 其中 PLi = P3L1 + PiD2 + P2D + P3D5 + P4D4 + P5D2 , PL2=PiDi + P2D2 + P3D3 + P4D2 + P5Dl。 一种能够高速产生调频网点的装置 1 , 如图 7所示, 该装置包括误差行存储 器 2 , 误差緩冲寄存器堆 3, 阐值比较电路 4 , 误差生成电路 5 , 误差緩冲寄存器 堆控制电路 6 , 灰度生成电路 7 , 误差行控制电路 8共七部分。
在上述装置中, 灰度生成电路 7根据当前处理像素和误差行存储器 2及误差 緩冲寄存器堆 3的内容得出最终的像素灰度, 然后通过阈值比较电路 4产生网 点, 同时最终的像素灰度还通过误差生成电路 5得出对应该像素的误差分配值, 然后通过误差缓冲寄存器堆控制电路 6和误差行控制电路 8来分别完成误差緩冲 寄存器堆 3和误差行存储器 2的一次更新。 '
在上述装置中, 误差行存储器 2是按列组织的存储器, 用来存储误差累积 值 , 它连接到误差行控制电路 8并由误差行控制电路 8控制;
误差緩冲寄存器堆 3由硬件触发器组成, 用来緩冲中间误差分配值, 它连接 到误差緩冲寄存器堆控制电路 6并由误差緩冲寄存器堆控制电路 6来控制; 灰度生成电路 7用来产生源像素的最终灰度, 它的输入连接到源像素和误差 緩冲寄存器堆控制电路 6 , 它的输出则提供给阁值比较电路 4和误差生成电路 5; 附图 8是本发明所述的一种能够高速产生调频网点的装置中灰度生成电路的 框图, 它由一个加法器产生最终像素的灰度值, 其中最终像素的灰度值等于源像 素的值加上从误差行中读出的对应该位置的当前误差值再加上误差緩冲寄存器堆 中緩冲的当前像素同一行的以前像素对当前像素的当前误差緩冲值, 以实现附图 2所示的 Stucki算法为例, 该电路的逻辑表达式为: 最终像素灰度=源像素值 + Lo+ ( P5D5+P4D3 ) 。
阈值比较电路 4用来将灰度生成电路 Ί所产生的灰度与阈值进行比较产生网 点, 它的输入连接到灰度生成电路 7 , 输出即是最终网点;
误差生成电路 5用来将由灰度生成电路 7所产生的灰度通过计算或查表而产 生当前像素的误差分配值, 它的输入连接到灰度生成电路 7 , 输出将提供给误差 緩冲寄存器堆控制电路 6和误差行控制电路 8;
误差緩冲寄存器堆控制电路 6 用来控制误差緩冲寄存器堆的更新, 它的输 入连接到误差生成电路 5和误差行控制电路 8及误差緩冲寄存器堆 3 , 它的输 出连接到误差行控制电路 8及误差緩冲寄存器堆 3和灰度生成电路 7 ; 附图 10 是本发明所述的一种能够高速产生调频网点的装置中误差緩冲寄存器堆控制电 路中多路器的框图, 误差緩冲寄存器堆控制电路通过 n 个多路器 ( mul t iplexer ) 来完成在每一像素处理周期内误差緩冲寄存器堆的一次更新 及每一行开始时的初始化清零, 其中 n为误差緩冲寄存器堆中的寄存器个数, 即对应误差緩冲寄存器堆中的每一个寄存器都包含有结构相同的一个多路器, 其中多路器的输入包括当前处理像素的所有误差分配值即当前误差分配值, 从 误差行中读出当前处理像素对应的下一行当前误差累积值及误差緩冲寄存器堆 中的原值, 多路器的输出即误差緩冲寄存器堆中寄存器的更新值, 以实现附图 2 所示的 S tuck i 算法为例, 当得出当前误差分配值 Dj 及当前误差累积值 Li 后, 该多路器的输出可以表示为对应于 PiDj , 多路器输出为 Pi + iDj , 对应于 P5Dj多路器输出为 Dj , ( i = 1. . 4 , j=1. . 5 ) , 对应于 PiLi, 多路器输出为 Pi + 1L1 , (i=3, 4) , 对应于 P5L1 , 多路器输出为 Li。
误差行控制电路 8用来控制误差行存储器 1 的读写, 它的输入来自误差生 成电路 5 和误差緩冲寄存器堆控制电路 6 , 它的输出连接到灰度生成电路 7 , 同时它还和误差行存储器 2相连。 附图 9是本发明所述的一种能够高速产生调 频网点的装置中误差行控制电路的框图, 它由读地址寄存器, 写地址寄存器, 读写控制电路及写入数据生成电路器组成, 其中读地址寄存器, 写地址寄存器 由当前处理像素坐标生成, 写入数据生成电路则由加法器组成, 它由原误累积 差值, 当前误差分配值及相关误差緩冲值得出一列误差行写入数据, 读写控制 电路则产生误差行存储器的读写控制信号, 以实现附图 2 所示的 Stuck i 算法 为例, 如当前处理像素列号为 j , 则读地址寄存器 = 4* j , 写地址寄存器则为 4* (j-2),误差行写入数据生成电路产生一列写入数据 PLi和 PL2 , 其中 PLi可表 示为 PLl = P3Ll + PlD2 + P2D4 + P3D5 + P4D4 + P5D2 , P 可表示为 PL2=PlDl + P2D2 + P 3D3 + P4D2 + P5Dl。 附图 11 是本发明所述的方法在处理一个像素时误差行存储器的读写过程 示意图。 以实现附图 2所示的 Stucki算法为例, 设 *所示为当前处理像素, 如 采用现有技术, 处理当前处理像素 *时需要读出附图 11 所示误差行中 PL2和 PLi, Lo和 Li及所有标记为 X的位置的原内容, 当前像素处理完毕后, 还需要 更新当前像素相邻的 12 个位置上的累积值, 而采用本发明所述的方法和装 置, 当处理当前像素 *时, 先从误差行存储器中读出 Lo和 Li , 其中 Lo用来产生 最终灰度, 而 Li被緩冲, 在以后用来产生 PLi , 当当前处理像素的误差分配值 Dj产生后, 可以得知误差行存储器中 PL2和 PLi的值可以被确定并写入, 在该 行以后的像素处理中误差行中该位置的内容不会改变, 同时因为影响 PL2和 PLi最终累积值的所有相关误差值被緩冲在误差緩冲寄存器堆中, 所以不需要 再次读取存储器即可完成 PL2和 PLi的写入, 而且因为本发明中误差行是按列 组织的, 因此读 LQ和 Li只需一次存储器读操作而写 PL2和 PLi也只需一次存 储器写操作, 可见采用本发明所述方法, 每处理一个像素时, 只需要一次存储 器读来读出 Li和 Lo及一次存储器写操作写入 PL2和 PLi即可。 从而网点产生的 速度得以大幅提高。 以上所述仅为本发明的其中一个实施例而已, 并不用以限制本发明, 如以上 用将误差扩散到 12个相邻位置的 Stucki 算法来描述本发明实施例, 但本发明同 样适用于基于误差扩散原理的, 将误差扩散到更多的点的网点生成方法, 因而凡 在本发明的精神和原则之内, 所作的任何修改, 等同替换, 改进等, 均应包含在 本发明的保护范围之内。

Claims

权 利 要 求 书
1.一种能够高速产生调频网点的方法, 包括以下步骤:
a )按列组织并初始化误差行存储器;
b )设定误差緩冲寄存器堆;
c )获取源像素;
d )依据当前源像素及误差行和误差緩冲寄存器堆的内容产生网点; e )更新误差行及误差緩冲寄存器堆;
' f) 重复步骤 c, d, e直到一行结束;
g) 重复步骤 b, c, d, e , f 直到图像结束。
2. 根据权利要求 1所述的一种能够高速产生调频网点的方法, 其特征在于: 步骤 a中按列组织并初始化误差行存储器时采用如下方法, 具体包括如下步骤: ·
1 )误差行的行数为 2的 n次方, 并且误差行存储器的总线宽度能够保证一次 可以从误差行存储器中读写 1 的 n次方个单元, 同时误差行的行数大于等于一个 像素的误差所能够扩散到的行数, 误差行循环重复使用, 上述的 n为正整数;
2 )列号相同的每 2的 n次方列误差位置相邻排列;
3 )误差行内容被初始化为 0。
3. 根据权利要求 1所述的一种能够高速产生调频网点的方法, 其特征在于: 步骤 b中设定误差緩冲寄存器堆时采用如下方法, 具体包括如下步驟:
1 )误差緩冲寄存器堆包括误差分配緩冲寄存器堆和误差累积緩冲寄存器堆两 部分, 其中误差分配緩冲寄存器堆的大小能够保存误差行某一位置的所有相关像 素所有误差分配值, 其中当前相关像素为当前处理行内误差值能扩散到这一误差 行位置所有像素, 误差累积緩冲寄存器堆的大小为足够保存读出的当前处理行的 下面行的原误差行中的所有相关误差累积值, 其中相关误差累积值为从读出误差 行某一位置上原误差累积值到该位置新的误差累积值被写入过程中所有读出的误 差行中的原误差累积值;
2 )初始化误差緩冲寄存器堆即每行处理开始时误差緩冲寄存器堆内容被初始 化为 0。
4. 根据权利要求 1所述的一种能够高速产生调频网点的方法, 其特征在于: 步骤 d 中采用依据当前源像素及误差行和误差緩冲寄存器堆的内容产生网点时采 用如下方法, 包括如下步骤:
1 )从误差行存储器中一次读出对应当前源像素位置及下面行同一列位置上的 误差累积值; 2 ) 由误差緩冲寄存器堆的内容计算出当前源像素的同一行的所有以前的像素 处理所产生的误差分配到当前源像素的误差值的和 S;
3)将当前像素对应位置上的原误差累积值加上 S, 再和当前源像素值相加得 出最终像素值, 将该值和阐值比较产生最终网点。
5. 根据权利要求 1所述的一种能够高速产生调频网点的方法, 其特征在于: 步骤 e中更新误差行及误差緩冲寄存器堆时采用如下方法, &括如下步骤:
1 )根据最终像素值和阈值得出当前像素产生的误差分配到所有相邻像素的值
Dj;
2) 误差分配緩冲寄存器堆中每一像素的误差分配值更新为后一像素的对应 值, 而最后一个像素的误差分配值更新为 Dj;
3 )误差累积緩冲寄存器堆中每一像素的误差累积值更新为后一像素的对应 值, 而最后一个像素的误差累积值更新为当前像素对应位置下面行上的原误差累 积值;
4 )根据误差緩冲寄存器堆内容得出误差行的一列的所有写入数据值并一次写 入误差行存储器中, 其中误差行一列中某一点的写入数据为存放在误差緩冲寄存 器堆中的扩散到该点的所有误差分配值和该点原误差累积值之和。
6. 一种实现如权利要求 1所述方法的能够高速产生调频网点的装置, 该装置 (1) 包括阈值比较电路(4) , 误差生成电路(5) , 其特征在于还包括: 误差行 存储器(2) , 误差緩冲寄存器堆(3) , 误差緩冲寄存器堆控制电路(6) , 灰度 生成电路(7) , 误差行控制电路(8) ;
所述误差行存储器(2)用来保存处理过程中的误差扩散结果; 误差緩冲寄存 器堆(3)用来緩冲中间误差值; 灰度生成电路(7)连接到源像素、 误差行存储 器(2)和误差緩冲寄存器堆(3) , 用来产生源像素经过误差扩散后的最终灰 度;
所述阈值比较电路(4)连接到灰度生成电路(7)将阈值和最终灰度比较产 生最终网点; 误差生成电路(5)连接到灰度生成电路(7) 由阈值和最终灰度计 算出当 '前误差分配值; 误差緩冲寄存器堆控制电路 (6) 来控制误差緩冲寄存器堆 的更新及输出, 它的输入来自误差生成电路( 5 )和误差行控制电路( 8 )和误差 緩冲寄存器堆( 3 ) 的内容反馈; 误差行控制电路( 8 )连接到误差行存储器( 2 ) 用来产生误差行存储器的读写地址及读写控制信号及读写数据, 它的输入来自包 括当前处理象素的坐标及误差緩冲寄存器堆控制电路(6)及误差生成电路
(5) 。
7. 根据权利要求 6所述的一种能够高速产生调频网点的装置, 其特征在于: 误差緩冲寄存器堆(3 )采用由触发器组成的寄存器堆实现, 它包括误差分配 緩冲寄存器堆和误差累积緩冲寄存器堆两部分, 其中误差分配緩冲寄存器堆大小 能够保存误差行某一位置的所有相关像素所有误差分配值, 其中当前相关像素为 当前处理行内误差值能扩散到这一误差行位置所有像素; 误差累积緩冲寄存器堆 大小为足够保存读出的当前处理行的下面行的原误差行中的所有相关误差累积 值, 其中相关误差累积值为从读出误差行某一位置上原误差累积值到该位置新的 误差累积值被写入过程中所有读出的误差行中的原误差累积值。
8. 根据权利要求 6所述的一种能够高速产生调频网点的装置, 其特征在 于: 灰度生成电路(7 ) 由一个加法器组成, 加法器的输入来自源像素、 误差行 中读出的当前误差值和误差缓冲寄存器堆中的当前误差緩冲值。
9. 根据权利要求 6 所述的一种能够高速产生调频网点的装置, 其特征在 于: 误差緩冲寄存器堆控制电路(6 ) 由 n个多路器组成, 其中 n为误差緩冲寄 存器堆中的寄存器个数, 多路器的输入包括当前误差分配值, 当前误差累积值 及误差緩冲寄存器堆中的原值, 多路器的输出为误差緩冲寄存器堆中寄存器的 更新值。
10. 根据权利要求 6所述的一种能够高速产生调频网点的装置, 其特征在 于: 误差行控制电路(8 ) 由读地址寄存器, 写地址寄存器, 读写控制电路及写 入数据生成电路组成, 其中读地址寄存器和写地址寄存器由当前像素的列号产 生, 写入数据生成电路则由加法器组成, 其中加法器的输入来自于原误差累积 值、 当前误差分配值及相关误差緩冲值。
PCT/CN2006/000567 2005-04-07 2006-03-31 Procede et appareil capable de production de points grilles de modulation de frequence a grande vitesse WO2006105721A1 (fr)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP06722219.0A EP1874031B1 (en) 2005-04-07 2006-03-31 Method and apparatus capable of producing fm lattice points in high speed
JP2008504606A JP4481343B2 (ja) 2005-04-07 2006-03-31 高速でfm格子点の発生を可能にする方法および装置
US11/918,077 US8363278B2 (en) 2005-04-07 2006-03-31 Method and apparatus capable of producing FM halftone dots in high speed

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN200510063255.3 2005-04-07
CNB2005100632553A CN1326384C (zh) 2005-04-07 2005-04-07 一种产生调频网点的方法和装置

Publications (1)

Publication Number Publication Date
WO2006105721A1 true WO2006105721A1 (fr) 2006-10-12

Family

ID=35038928

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2006/000567 WO2006105721A1 (fr) 2005-04-07 2006-03-31 Procede et appareil capable de production de points grilles de modulation de frequence a grande vitesse

Country Status (5)

Country Link
US (1) US8363278B2 (zh)
EP (1) EP1874031B1 (zh)
JP (1) JP4481343B2 (zh)
CN (1) CN1326384C (zh)
WO (1) WO2006105721A1 (zh)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100435548C (zh) * 2006-09-15 2008-11-19 北京大学 一种能够同时产生多位调频网点的方法及装置
CN1964423A (zh) 2006-11-07 2007-05-16 北京大学 采用双份误差行存储器产生图像网点的方法和装置
EP2159709B1 (en) * 2007-06-15 2013-01-02 Fujitsu Limited Error correcting method and computing element
US9007659B1 (en) * 2013-12-16 2015-04-14 Xerox Corporation Re-ordered error diffusion for fast implementation
CN110910305B (zh) * 2019-11-25 2024-03-15 北京京隽科技有限公司 灰度图像半色调方法和装置、设备及存储介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11157132A (ja) * 1997-11-27 1999-06-15 Dainippon Printing Co Ltd デジタル階調画像の印刷方法および階調画像が表現された印刷物
US20010021275A1 (en) * 2000-03-10 2001-09-13 Fujitsu Limited Image processing method, image processor, and storage medium thereof
US20040130753A1 (en) * 2003-01-06 2004-07-08 Crounse Kenneth R. Halftone method and system using hybrid AM/FM screening for highlight/shadow tonal regions

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5337160A (en) 1992-07-01 1994-08-09 Hewlett-Packard Error diffusion processor and method for converting a grey scale pixel image to a binary value pixel image
US5917614A (en) * 1992-11-30 1999-06-29 Levien; Raphael L Method and apparatus for error diffusion screening of images with improved smoothness in highlight and shadow regions
JPH07210676A (ja) * 1994-01-27 1995-08-11 Canon Inc 画像処理装置
JP3290867B2 (ja) 1995-11-07 2002-06-10 ブラザー工業株式会社 多値データ変換装置及び多値データ変換方法
US5974228A (en) * 1997-01-28 1999-10-26 Hewlett-Packard Company Image rendition by plural-row error diffusion, for faster operation and smaller integrated circuits
US5946455A (en) * 1997-10-02 1999-08-31 International Business Machines Corporation Model based error diffusion with correction propagation
US6014227A (en) * 1998-04-30 2000-01-11 Hewlett-Packard Co. Printer with progressive column error diffusion system and method of using same for improved printer throughput
JP4335467B2 (ja) * 2000-03-07 2009-09-30 セイコーインスツル株式会社 濃淡画像の階調再現方法および装置
JP3990860B2 (ja) 2000-09-20 2007-10-17 キヤノン株式会社 画像処理装置及び画像処理方法
US7081972B2 (en) * 2000-09-20 2006-07-25 Canon Kabushiki Kaisha Image processing apparatus and image processing method
JP2002185788A (ja) 2000-10-06 2002-06-28 Seiko Epson Corp 画像処理装置、画像処理方法、印刷制御装置、および記録媒体
JP4031240B2 (ja) * 2001-12-20 2008-01-09 株式会社リコー 画像処理装置
CN1144153C (zh) * 2002-12-30 2004-03-31 北京北大方正电子有限公司 一种改善调频网中间调质量的数值的调频挂网方法
US7620255B2 (en) * 2005-06-24 2009-11-17 Seiko Epson Corporation Image processor and image processing program for binary processing

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH11157132A (ja) * 1997-11-27 1999-06-15 Dainippon Printing Co Ltd デジタル階調画像の印刷方法および階調画像が表現された印刷物
US20010021275A1 (en) * 2000-03-10 2001-09-13 Fujitsu Limited Image processing method, image processor, and storage medium thereof
US20040130753A1 (en) * 2003-01-06 2004-07-08 Crounse Kenneth R. Halftone method and system using hybrid AM/FM screening for highlight/shadow tonal regions

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP1874031A4 *

Also Published As

Publication number Publication date
US20090051972A1 (en) 2009-02-26
US8363278B2 (en) 2013-01-29
CN1326384C (zh) 2007-07-11
EP1874031B1 (en) 2016-07-27
JP4481343B2 (ja) 2010-06-16
CN1668062A (zh) 2005-09-14
EP1874031A1 (en) 2008-01-02
EP1874031A4 (en) 2008-11-26
JP2008535419A (ja) 2008-08-28

Similar Documents

Publication Publication Date Title
JP3356612B2 (ja) 高速な輪郭スムージング方法及び装置
US6064407A (en) Method and apparatus for tiling a block of image data
CN112767983B (zh) 刷新控制电路及存储器
WO2006105721A1 (fr) Procede et appareil capable de production de points grilles de modulation de frequence a grande vitesse
WO2022206556A1 (zh) 图像数据的矩阵运算方法、装置、设备及存储介质
JPH0698159A (ja) 2値化方法及び装置
Kaukoranta et al. Vector quantization by lazy pairwise nearest neighbor method
CN108876701A (zh) 一种基于游程的单次扫描通域标记方法及其硬件结构
JP5113174B2 (ja) マルチサイトfmスクリーンドットを同時に生成する方法及び装置
CN1937701A (zh) 图像处理装置及其处理方法
US8245091B2 (en) Methods and devices for generating dots of an image by using two error row memories
JP2004164549A (ja) 画像処理装置、描画処理方法及び該方法を実行するためのプログラム
JP3215598B2 (ja) 最小誤差記憶装置
US20230168809A1 (en) Intelligence processor device and method for reducing memory bandwidth
CN101867681A (zh) 一种数字图像处理方法及系统
JP2861435B2 (ja) パイプライン形演算装置
US8347069B2 (en) Information processing device, information processing method and computer readable medium for determining a processing sequence of processing elements
JPS63266560A (ja) バツフア記憶制御方式
JP2006050299A (ja) ディザリング装置
Regan Linear-Time Algorithms in Memory Hierarchies.
JP2005018693A (ja) 画像処理方法
JPS6217790A (ja) 文字パタ−ン発生回路
JPH0531790B2 (zh)
JPH07334480A (ja) 遺伝的最適化方法
JP2001313829A (ja) 画像処理方法及び装置

Legal Events

Date Code Title Description
DPE2 Request for preliminary examination filed before expiration of 19th month from priority date (pct application filed from 20040101)
121 Ep: the epo has been informed by wipo that ep was designated in this application
NENP Non-entry into the national phase

Ref country code: DE

WWE Wipo information: entry into national phase

Ref document number: 11918077

Country of ref document: US

Ref document number: 2008504606

Country of ref document: JP

WWW Wipo information: withdrawn in national office

Ref document number: DE

REEP Request for entry into the european phase

Ref document number: 2006722219

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 2006722219

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: RU

WWW Wipo information: withdrawn in national office

Ref document number: RU

WWP Wipo information: published in national office

Ref document number: 2006722219

Country of ref document: EP