JP3555005B2 - Image reduction conversion apparatus and method, and recording medium storing the program - Google Patents

Description

【００２７】
として求める。なお、一例として、ｇ（ｓ，ｔ）＝１として
【００２８】
【数２】

【００２９】
とすることが可能である。これにより、加算値ｒｉｊは、０〜Ａ×ｎ×ｎまでの値を得ることができる。これが変換画素の値を算出する元になる。加算値ｒｉｊは、変換画像を原画像へ写像した場合の各小領域の変換画素について求める。
【００３０】
最大値算出手段１３４は、各小領域内での加算値ｒｉｊの最大値を算出する。これは、加算値決定手段１３３で求めたｒｉｊを記憶しておき、該当小領域の中から最大値を算出することで可能である。なお、本発明は、小領域の中からの最大値でなく、分類された領域内の中からの最大値をもとめる手段で代替することが可能である。これは、小領域を用いないで領域を分類する方法を本発明に適用した場合には、小領域内でなく分類された領域内での最大値とすることにより、本発明が実現できる。なお、本発明での中間調の階調は、数値が大きくなるに従い、黒色濃度が増すという前提で最大値を求めるが、システムによっては数値が小さくなるに従い黒色濃度が増すものもあり、このような場合には最小値を求めることにより本発明を実現する。この最大値は、２値領域の黒濃度を保持しておくものであり、後述する。
【００３１】
参照画素加算値算出手段１３０で求まった各変換画素の参照画素加算値２４０はメモリ装置２００に格納される。この参照画素加算値２４０には当該小領域中での加算値の最大値が付加され、また、領域分類手段１２０で分類された当該小領域が中間調領域か２値領域かを示す識別子（中間調２値識別子）も属性として付加される。
【００３２】
〔変換画素値算出手段１４０〕
変換画素値算出手段１４０は、図５に示すように、中間調領域の変換画素値算出手段１４１と２値領域の変換画素値算出手段１４２の２つからなる。中間調領域の変換画素値算出手段１４１は、参照画素加算値２４０を入力し、属性として付加されている中間調／２値識別子を参照して、領域分類手段１２０により中間調領域と分類されたときに処理が実施される。２値領域の変換画素値算出手段１４２は、領域分類手段１２０により２値領域と分類されたときに処理が実施される。
【００３３】
ここで、原画像の中間調の階調の最大値をＡとし、変換画像の中間調の階調の最大値をＢとする。また、白値を０とし、最大の黒色濃度はＡまたはＢとする。例えば、２５６階調ならば取りうる階調の値は０〜２５５であり、０が白、２５５が黒を示すことにする。また、１３４で算出した小領域内の最大値をＭとする。
【００３４】
ここで、該当の小領域が中間調領域と領域分類手段１２０で分類されているときには、中間調領域の変換画素値算出手段１４１により、変換画素Ｑｉｊの値を、
【００３５】
【数３】

【００３６】
として算出する。反対に、該当の小領域が２値領域と分類されているときには、２値領域の変換画素値算出手段１４２により、変換画素Ｑｉｊの値を、
【００３７】
【数４】

【００３８】
として算出する。なお、原画像が２値画像の時はＡ＝１とすることで、算出が可能である。このように、中間調領域の場合は加算値を原画階調数と参照画素個数で割り、一方、２値領域の場合は加算値を小領域内の最大値Ｍで割り変換画素の値とする。このとき、Ｍ≦Ａ×ｎ×ｎである。このＭは、小領域の中の最大値であり、最も濃度の高い値を保持するものである。中間調領域の変換画素の値はｒｉｊを（Ａ×ｎ×ｎ）、すなわち、（原画階調数×参照画素個数）で割るが、２値領域ではｒｉｊをＭで割ることにより、変換画素の値が薄くなるのを抑える効果を有している。これは、２値領域において、特に変換比率が小さくなるにつれて、加算する参照画素個数（ｎ×ｎ）個が増加し、加算値が平準化されるために、最も強い濃度の値が薄くなる現象が生じるからである。このため、２値領域において、小領域の中の加算値の最大値Ｍを求め、そのＭを最大黒濃度として（原画階調数×参照画素個数）と置き換えることにより、特に変換比率が小さい場合の、原画像の細い線が薄くなるのを防止できる大きな効果を有している。
【００３９】
変換画素値算出手段１４０の出力が変換画像２５０としてメモリ装置２５０に格納される。画像出力装置４００には、この変換画像２５０をそのま出力することも可能である。
【００４０】
〔変換画像修正手段１５０〕
変換画像修正手段１５０は、図６に示すように、中間調出力値決定手段１５１と２値化手段１５２からなる。
中間調出力値決定手段１５１は出力機器の表現できる階調数と変換画像の有する階調数が一致しない場合に、整合性をとる手段である。変換画像を５１２階調まで表現できる変換画として作成する場合、Ｂ＝５１１として０〜５１１の値を表現することができる。一度作成した変換画像を、出力機器にあわせて、表現する階調数を最終的に変化させるものである。例えば、４階調しか表現できない出力機器の場合など、２５６階調を４階調の限られた中間調値へ割り当てて、実際に出力する値を決定する。反対に、出力機器が２５６階調表現可能であり、画像が４階調のみの場合、白や黒の偏りなく、６４階調おきに平均的に分布させて表現した値を決定する。
【００４１】
なお、中間調出力値決定手段１５１は本発明に必須な手段ではないことがある。例えば、２５６階調表現の中間調画像をそのまま４階調表現の可能な出力機器へ入力すると、自動的に出力する階調数、４階調へ自動的に割り当てることができる出力機器がある。このような機器へ出力する場合には、本発明においては中間調出力値決定手段１５１は必要がない。これにより、本発明の請求範囲を限定することにはならない。
【００４２】
２値化手段１５２は、中間調領域と２値領域にそれぞれ対応したしきい値を設け、中間調表現の変換画像を２値化し、２値画像を出力する手段である。この処理は、次の通りである。一例として、縮小変換した中間調画像がＢ＝２５５、すなわち、２５６階調の画像を作成する。中間調領域と２値領域とに分けて、予め中間調領域用ｓ１と２値領域用のｓ２の値定めておく。例えば、ｓ１＝１２８とし、ｓ２＝６４とする。ｓ１とｓ２を記憶しておき、領域分割手段１２０により中間調域と分類されたときにはｓ１の２値化処理が実施され、領域分割手段１２０により２値領域と分類されたときはｓ２の２値化が実施される。これにより、縮小変換された中間調画像の２値化することが実現可能である。さらに、中間調領域と２値領域のしきい値を別々に設定することにより、中間調領域の最適なしきい値と、２値領域の最適なしきい値が設定でき、それぞれに最良の変換画品質が作成できる。
【００４３】
このように、２値化手段１５２により、本実施例による中間調画像を２値化して２値画像が生成できる。なお、２値化に関しては、１つの変換画素を対象に２値化する手段と、複数の変換画素を集めて２値化する手法があり、どちらも２値化手段１５２へ適用することが可能である。１つの変換画素のみを２値化するのは、処理が簡易で、ハードウェアが小さくなる利点を有している。複数の変換画素を集めて２値化する手法は、処理や装置が複雑になるが、特に写真などの中間調領域を２値化するときに階調特性を損なわないような従来手法が適用でき、品質劣化を抑えて２値化することができる利点を有している。
【００４４】
以上の実施例の説明においては、２値領域と中間調領域に分けた変換画素の値を決定する際に、中間調領域は加算値を参照画素個数と原画階調数で割る手法を、２値領域は加算値ｒｉｊを小領域内の最大値Ｍで割る手法を示した。これに対し、中間調領域の係数α１、２値領域の係数α２それぞれの係数をあらかじめ定めて記憶しておき、その領域ごとにα１とα２を用いて実現する本発明の別の実施例を以下に示す。ここで、小領域が中間調領域と領域分割手段１２０で分類されているときに、中間調領域の変換画素値算出手段１４１において、変換画素Ｑｉｊの値を、
【００４５】
【数５】

【００４６】
を算出して求める。また、小領域が２値領域と領域分類手段１２０で分類されているときに、２値領域の変換画素値算出手段１４２において、
【００４７】
【数６】

【００４８】
を算出して、求める。なお、原画像が２値画像の時はＡ＝１とする。
【００４９】
このように、小領域が中間調領域の場合は係数α１を掛け、２値領域の場合はα２を掛けて、変換画素の値を算出する。この二つの係数を用いることにより、最大値算出手段１３４、すなわち最大値Ｍを用いないで変換画を作成することができる。
【００５０】
これは、変換比率が特に小さくなると、２値領域の文字や細い線が薄くなるため、品質の劣化が大きくなる。このため、２値領域に対しては、係数α２を大きくし変換画素の濃度を黒色濃度を強調し、これにより生成される最適となる変換画の品質を生成するようにしたものである。反対に、中間調領域は、このような薄くなる現象がないために、α１を１に近い値としておく。すなわち、中間調領域の場合をα１、２値領域の場合をα２として、α１≦α２として、特に２値領域の変換画素のみを黒色濃度を強調することが本発明の大きな特徴である。ここで、係数α１とα２の具体的な例として、例えば、α１＝１，α２＝３と定める。このように、係数α１，α２を予め定義しておく。なお、α１，α２の値は一例であり、この値に固定するものでなく、出力機器の種別により最適な値とする事が可能である。
【００５１】
ここで説明したように、原画像の領域を２値領域と中間調領域に分け、２値領域の変換画素の階調値を強調するように係数α２を定め、かつ中間調領域の階調特性を悪化させないようにα１を定めるものであり、領域ごとに変換画素を決定する手段を有していることにより、画品質劣化を防ぐことができ、これによりきわめて大きな効果を有している。
【００５２】
以上説明したように、原画像を中間調領域と２値領域に分け、それぞれ変換画素の値を算出する手段を持つことが、本発明の特徴である。
【００５３】
さらに、中間調領域では参照画素の加算値を原画階調数と参照画素個数で割り、２値領域では加算値を小領域内の最大値Ｍで割る手段を持つことが本発明の特徴である。
【００５４】
また、小領域の最大値Ｍを求めることなく、中間調領域には係数α１、２値領域には係数α２を予め定めておき、加算値へそれぞれの係数を掛けたものを用いて変換画素の値を決定する別の実施手段を実現することが本発明の特徴である。
【００５５】
なお、本発明では、水平方向と垂直方向の変換比率が同一の場合について説明したが、水平方向と垂直方向の変換比率が異なる場合、及びその一方が拡大変換に対しても、適用可能である。
【００５６】
また、本発明では原画像および変換画像ともに、白黒の中間調画像もしくは白黒の２値画像である場合を対象に説明してきた。しかし、本発明は画像として白黒に限定されるものでない。白黒画像以外、例えば、白色地に赤文字のような２色で表現されている画像に対しても適用が可能である。すなわち、情報の有無を例えば中間調画像の最大値２５５、最小値０で表現すると、２５５を赤、０を白の下地として、本技術における中間調画像へと帰着できる。これによる縮小変換された中間調画像は、最大の階調を持つ色を赤とし、最低の階調を白地として、中間の濃度を赤の階調成分で表現することで、白赤の２色の画像を中間調画像へ縮小した画像が得られる。また、それを２値化処理した結果は、情報がある場所を赤で、ない場所を白地に彩色したことに相当する。
【００５７】
以上、本発明の一実施例を説明したが、図１の各手段１１０〜１５０、さらには図２乃至図６に示した具体的実現手段は、コンピュータプログラムを用いて実現することでもよい。その機能や手順、アルゴリズムをコンピュータで実行可能なように記述したプログラムは、フロッピーディスクやメモリカード、ＣＤ−ＲＯＭ等に記録して提供することも可能である。
【００５８】
【発明の効果】
本発明によれば、中間調画像を表現できる出力機器へ出力する場合、原画像を、中間調領域と２値領域とに領域分類を行い、それぞれの領域ごとに変換画素の濃度値を算出することで、最適な品質の中間調画像を作成して出力できる。
また、中間調領域では、参照画素の加算値を参照画素の個数と原画像の階調数で割り、２値領域では、加算値を小領域の最大値で割り、それぞれ変換画素の値を求めることにより、最適な変換画素の値をそれぞれ算出できる。
また、中間調領域には係数α１、２値領域には係数α２を設けて、それぞれ加算値へ係数を掛けることにより、最大値Ｍを用いなくても、それぞれの領域に最適な変換画素の値を求められる。
以上により、中間調領域でも、２値領域でも、それぞれ画品質劣化の少ない最良の変換画を作成することができる。特に、変換比率が小さくなったとき、さらには出力機器が高解像度になったときに、２値領域において変換画が薄く表現されて、変換画品質が著しく劣化してしまうという従来の欠点を本発明により克服できる。
【図面の簡単な説明】
【図１】本発明における一実施例のブロック図である。
【図２】原画像種別検出手段の構成例を示す図である。
【図３】領域分類手段の構成例を示す図である。
【図４】参照画素加算値算出手段の構成例を示す図である。
【図５】変換画素値算出手段の構成例を示す図である。
【図６】変換画像修正手段の構成例を示す図である。
【図７】原画像を小領域へ分割し、２値領域と中間調領域がある例を示す図である。
【図８】中間調領域と２値領域との領域分類を実現する手法の一例を示す図である。
【図９】原画素と変換画素との位置配置関係の具体例を示す図である。
【符号の説明】
１１０ 原画像種別検出手段
１２０ 領域分類手段
１３０ 参照画素加算値算出手段
１４０ 変換画素値算出手段
１５０ 変換画像修正手段[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image reduction conversion method and apparatus, and more particularly, to reducing and outputting an original image to an image output device or the like capable of expressing a halftone, when a halftone area and a binary area in the original image are reduced. Image reduction conversion to obtain a reduced image with extremely low image quality degradation by classifying and performing halftone expression conversion of optimal quality for each areaApparatus and method, and recording medium recording the programIt is about.
[0002]
[Prior art]
The conventional reduction conversion method is a method of obtaining a value of a halftone-expressed converted pixel by associating the position of the original pixel with the converted pixel, performing weighted addition according to the position information of the original pixel around the converted pixel. Is common. Furthermore, there is a method of performing weighting according to the area that contributes. In any case, the related art does not determine the conversion pixel in consideration of the type of the image included in the original image, but determines the conversion pixel uniformly.
[0003]
For this reason, when the original image includes a halftone area (an area such as a photograph) and a binary area (an area represented by a line drawing, an area of a character, an area of a document, etc.), Could not achieve reduced quality conversion. In particular, as the conversion ratio becomes smaller, the number of original pixels necessary for determining the value of the converted pixel increases, and the average is taken. Therefore, in a reduced converted image of a binary area, a thin line or a character becomes thinner. If the quality is greatly deteriorated, or if the quality of the binary area is suppressed, the gradation characteristic of the halftone area is deteriorated, the quality is deteriorated, and there is no means to overcome these.
[0004]
[Problems to be solved by the invention]
When reducing the original image, even if the conversion ratio is as small as 1/20 or a relatively large conversion ratio as large as 1/2, quality deterioration is minimized and the converted image of the best quality is displayed on the display. Or a request to print to a printer.
[0005]
SUMMARY OF THE INVENTION It is an object of the present invention to perform image reduction conversion on a source image including a halftone area and a binary area, regardless of the conversion ratio, to obtain a converted image with reduced image quality deterioration.Apparatus and method, and recording medium recording the programIs to provide. Especially, when the conversion ratio is small, image reduction conversion that can generate a converted image with significantly reduced quality deteriorationAn object of the present invention is to provide an apparatus and a method, and a recording medium on which the program is recorded.
[0006]
[Means for Solving the Problems]
In the present invention, the original image is divided into a plurality of small regions, each small region is classified into a halftone region or a binary region, and the value of the conversion pixel is determined according to the type of the classified region. The main feature is to decide. As a result, a converted image with the best image quality can always be generated in both the halftone region and the binary region regardless of the reduction ratio. In particular, as the conversion ratio becomes smaller, the effect of suppressing the quality deterioration is reduced. Have. Further, in the present invention, a function of dividing the generated converted image into a halftone area and a binary area of the original image and having a threshold value, and providing a function of generating a high-quality binary image by binarization is provided. It is characterized by.
[0007]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. Hereinafter, a case where the conversion ratios in the horizontal direction and the vertical direction are the same will be described. The pixels of the original image are called original pixels, and the pixels of the converted image are called converted pixels.
[0008]
FIG. 1 is a block diagram showing an embodiment of the present invention, in which an image processing apparatus 100 comprising a CPU and a built-in memory constituting a reduction conversion apparatus, and a hard disk for storing original images, converted images, in-process images, parameters, and the like. Memory devices such as200,processingTargetThe apparatus includes an image scanner for inputting an image, another image input device 300, and an image output device 400 such as a display or a printer for outputting a converted image or the like. Here, the image processing apparatus 100 functionally detects whether the original image is a halftone image or a binary image, and converts the binary image into a halftone image. The converted halftone image) is divided into small areas, and the area classifying means 120 classifies the converted image into a halftone area and a binary area, respectively, detects a converted pixel position of the converted image, obtains a reference pixel of a conversion pixel of interest, and calculates the value. Reference pixel addition value calculation means 130 for adding, conversion pixel value calculation means 140 for calculating the values of the conversion pixels in the intermediate area and the binary area to generate a conversion image, and having the number of gradations and the image that can be expressed by the halftone output device Conversion image correcting means for obtaining consistency when the number of tones does not match, determining a value to be output, and binarizing the halftone image by a threshold value corresponding to the classification of the area to generate a binary image It is roughly divided into 150.
[0009]
It should be noted that a part of the original image type detecting unit 110 and the converted image correcting unit 150 may not be essential to the present invention. This is because, as will be described later, in the configuration of the present embodiment, it may be possible to operate without any problem even without a certain means.
[0010]
The operation of each means 110 to 150 in the embodiment of FIG. 1 will be described below. The conversion ratio is 1/4 in both the horizontal and vertical directions. It is also assumed that the original image is input by the image input device 300 and is stored in the memory device 200.
[0011]
[Original image type detecting means 110]
The original image type detection unit 110 is a unit that performs preprocessing so that the present invention can be applied to both a binary image and a halftone image. FIG. 2 shows a configuration example of the original image type detection means 110.
[0012]
The original image determining means 111 receives the original image 210, determines whether the image is a binary image or a halftone image, and outputs it as it is if it is a halftone image, or outputs it as a halftone image if it is a binary image. Hand over to 112. Here, a binary image or a halftone image is used to mean whether one pixel is represented by one bit or not, and a pseudo halftone that expresses a photograph or the like in black and white density is a binary image. To determine whether the original image is a binary image or a halftone image, (1) If the original image has an identifier such as a binary image or a halftone image as an input file format, look at the identifier and look at the binary image. (2) The effective information amount of each pixel is determined. If the effective information amount is 1 bit, the image is determined to be a binary image, and the effective information amount of each pixel is a plurality of bits. Then, a halftone image and a judgment method can be applied.
[0013]
The halftone image conversion unit 112 performs conversion to a halftone image when the original image 210 is a binary image. This is because, in a binary image represented by 0 (white) and 1 (black), 1 of the binary image is replaced with the maximum value of the halftone image, for example, 255 for 256 gradations, and 0 of the binary image is replaced. Is replaced with 0 of the halftone image, thereby replacing the image with only the values of 0 (white) and 255 (black) of the halftone image. In this way, even if the original image is a binary image, reduction conversion can be realized as a halftone image.
[0014]
The output image of the original image type detecting means 110 is stored in the memory device 200 as the converted halftone image 220. The original image type detecting means110May not be an essential means of the present invention. This can be used, for example, in a system with only a scanner where the original image has 256 gradations.Be doneIn this case, the original image can be fixed at 256 gradations, and even without the original image type detecting means 110, the following means 120 to 150 can operate without any trouble and realize the effects of the present invention. That's why.
[0015]
[Area classification means 120]
The area classifying means 120 determines in advance that the present invention changes the algorithm for determining the value of the conversion pixel between the halftone area and the binary area to create a converted image with little image quality deterioration. (Tone image) is divided into a halftone area and a binary area. FIG. 3 shows a configuration example of the area classification means 120.
[0016]
The small area dividing means 121 receives the converted halftone image 220 and divides it into a plurality of small areas. An algorithm for dividing the converted halftone image (original image) into a plurality of small areas will be described using a specific example of division shown in FIG. In the original image shown in FIG. 7, a photograph is attached to a hatched area (hatched area) in the upper right, and the rest is a document. First, the number of small areas divided into E in the horizontal direction and F in the vertical direction is set to a predetermined number. By defining the condition of E = 6 in the horizontal direction and F = 6 in the vertical direction, equal, it is possible to divide as shown in FIG. Further, by increasing the number of E and F, fine division is possible. For example, as shown in the example of FIG. 7, in the case of a rough division such as E = F = 6, when the range of a photograph is elliptical, the range may not be properly classified as a halftone area. On the other hand, by increasing the number of divisions, such as E = 10 and F = 20, fine division can be performed, and it can be more flexibly adapted to the configuration of the original photograph and the document. Original image. On the other hand, when all the original images are documents and when all the images are photographs, the application is limited. If E = 2 and F = 1, or extremely, E = F = 1 Since the classification processing is not performed in detail, high-speed processing without division can be realized.
[0017]
The small area classifying unit 122 is a unit that determines whether the divided small area is classified into a halftone area or a binary area. FIG. 8 shows a specific explanatory diagram of the algorithm of the small area classification means 122. For example, three 5 × 5 original pixel patterns are extracted from the small area, the number of horizontal conversion points and the number of vertical change points are checked, and both the horizontal and vertical change points are specified values (for example, 2). If there are two of the three cases, the halftone region is determined. If the number is less than three, the region is determined to be the binary region. This is repeated for each small area. As a result, the original image can be classified into two regions, a halftone region and a binary region. In this way, in the original image of FIG. 7, each small area of the hatched portion can be determined as a halftone area, and the other small areas can be determined as binary areas.
[0018]
The halftone / binary pattern storage unit 123 is a unit that stores all possible patterns of the original image pattern. Specifically, a halftone pattern for determining a halftone area and a binary pattern for determining a binary area are stored in advance. For example, all possible states of the 5 × 5 original pixel pattern shown in FIG. 8 that can be taken as a black and white pattern are stored, and the small area dividing means 122 can check the coincidence with all the patterns. Thus, the small area classification unit 122 can realize the same function as the classification function based on the detection of the change point of the 5 × 5 original pixel pattern. In addition, in all cases where the original pixel pattern shown in FIG. 8 can be taken, there are 2 25 powers. By recording all of them, all black and white patterns can be classified. , It is possible to classify the halftone area and the binary area with higher accuracy than by detecting the change point. It is to be noted that a method equivalent to recording all 2 25 powers, that is, those which can be reduced to the same pattern can be compressed and stored only as necessary patterns, thereby greatly reducing the storage capacity. Can be.
[0019]
The small area classifying means 122 adds an identifier (halftone / binary identifier) indicating to each small area of the original image (converted halftone image) whether the small area belongs to the halftone area or the binary area. , Is stored in the memory device 200 as the halftone / binary classification image 230. For example, in the original image of FIG. 7, a halftone identifier is added to each small area of the hatched portion, and a binary identifier is added to each other small area. This identifier will be used by the conversion pixel value calculation means 140, which will be described later, to sort the conversion pixel value determination algorithm for each halftone area or binary area.
[0020]
The means for classifying the halftone area and the binary area is not limited to the configuration shown in FIG. This is because the present invention has means for classifying the halftone area and the binary area of the original image and determining the values of the converted pixels, respectively, so long as the image quality deterioration can be suppressed. Other conventional methods can be applied to the method for classifying the binary region. As an example of another area classification method, “Binarized expression of mixed image of characters / dots / photographs—continuous adaptive binarization using image area classification variables—: 20, 5, PP476-483, 1991 , And a region segmentation method (Akiyama, Masuda: “Area segmentation of document using both peripheral distribution, line density, and circumscribed rectangle feature”), IEICE Trans. (D), J-69-D, No. 8 (1986/8) )) And an area extraction method (Hase and Hoshino: “Document Image Area Extraction Method Using Two-Dimensional Fourier Transform”, IEICE (D), J-67-D, No. 9 (1984/9)). No. When these are used, the halftone area and the binary area can be classified without the small area dividing means 121 and the halftone / binary pattern storage means 123, and the following means 130 to 150 operate without any trouble. In addition, the effects of the present invention can be realized. Further, for an original image of a specific in-house document in which the positions of the halftone area and the binary area are fixedly determined, it is possible to determine whether the halftone area or the binary area when determining the conversion pixel. Even without the small area classification means 122 and the pattern storage means 123, the present system can be constructed, and the effects of the present invention can be realized.
[0021]
[Reference pixel addition value calculation means 130]
The addition value calculation means 130 is a means for obtaining an addition value of reference pixels around the conversion pixel of interest before calculating the value of the conversion pixel in the conversion image. FIG. 4 shows a configuration example of the reference pixel addition value calculation means 130.
[0022]
The converted pixel position detecting means 131 obtains the positional relationship between the converted pixels and the original pixels when the converted image is mapped on the original image (here, the halftone / binary classified image). FIG. 9 is a diagram illustrating a case where the converted image is mapped to the original image when the conversion ratios in the horizontal direction and the vertical direction are both １ ／. In the conversion pixel position detecting means 131, as shown in FIG. 9, the conversion image is mapped to the original image, and the relative position of the conversion pixel of interest from the original pixel closest to the upper left is obtained as the position of the conversion pixel of interest on the original image. . For example, for the converted pixel Q11, the original pixel P22 is the original pixel closest to the upper left, and for the converted pixel Q21, the original pixel P62 is the original pixel closest to the upper left.
[0023]
Next, the reference pixel determination unit 132 obtains a plurality of original pixels around the converted pixel as reference pixels. In this method, n necessary horizontal pixels × n vertical pixels of a plurality of necessary original pixels are obtained with the conversion pixel of interest as the center, and this is set as a reference pixel. Note that the reference pixels are a plurality of original pixels necessary for determining the value of the conversion pixel. Here, the number n is obtained by taking the value of the reciprocal of the conversion ratio as a natural number. For example, if the conversion ratio is 1/4, n = 4, and if the conversion ratio is 1 / 4.2, n = 5. In the specific example of FIG. 9, 4 × 4 = 16 around Q11 is obtained, and the results are P11 to P44. In the case of the next conversion pixel Q21, 4 × 4 = 16 pixels around the pixel Q21, and P51 to P84. Further, Q12 is 16 pieces of P15 to P48, and Q22 is 16 pieces of P55 to P88. The original pixels surrounded by each line are set as reference pixels.
[0024]
In the description of the present embodiment, as an example, n × n = 4 × 4 as n = 4 when the conversion ratio is ４. However, the number is not limited to n × n, and a method using (n−1) × (n−1) = 9 is also possible, and a method using (n + 1) × (n + 1) = 25 is also possible. is there. Further, it is also possible to use, as the reference pixel of Q11, a pixel that does not use one of the 16 pixels of P11 to P44 as the reference pixel. Further, 16 pixels of P11 to P44 can be used as P51, P52, P53, P54 It is also possible to add four. That is, the number of reference pixels is not limited to one. It is possible to increase or decrease the number slightly based on the fact that the number of pixels in the horizontal direction × vertical direction based on the reciprocal of the conversion ratio is used as the reference pixel with the conversion pixel as the center. In the following description, the number of reference pixels required for obtaining a converted pixel is generally described as (n × n).
[0025]
The addition value determination unit 133 is a unit that obtains the addition value of the reference pixel of the conversion pixel of interest, and performs the following processing. First, the original pixel obtained by the reference pixel determining means 132
(Reference pixel) Pst is read. Next, the values of the reference pixels are added. Here, the original imageHalftoneAssume that the maximum value of the gradation is A, the white value is 0, and the black value is A. For example, in the case of 256 gradations, possible gradation values are 0 to 255, where 0 indicates white and 255 indicates black. Also, the maximum value of the halftone gradation of the halftone image to be output is B. At this time, assuming that a function indicating position information between the reference pixel and the conversion pixel of interest is g (s, t), an added value rij to Qij is
[0026]
(Equation 1)

[0027]
Asking. As an example, assuming that g (s, t) = 1
[0028]
(Equation 2)

[0029]
It is possible to As a result, the added value rij can have a value from 0 to A × n × n. This is the basis for calculating the value of the converted pixel. The addition value rij is obtained for the conversion pixel of each small area when the conversion image is mapped to the original image.
[0030]
The maximum value calculation means 134 calculates the maximum value of the added value rij in each small area. This can be achieved by storing rij obtained by the addition value determination means 133 and calculating the maximum value from the corresponding small area. It should be noted that the present invention can be replaced by means for determining the maximum value from within the classified area instead of the maximum value from within the small area. This is because when the method of classifying an area without using a small area is applied to the present invention, the present invention can be realized by setting the maximum value not in the small area but in the classified area. In the present invention, the halftone gradation is determined to have a maximum value on the assumption that the black density increases as the numerical value increases.However, depending on the system, the black density increases as the numerical value decreases. In such a case, the present invention is realized by finding the minimum value. This maximum value holds the black density of the binary area and will be described later.
[0031]
Reference pixelThe reference pixel addition value 240 of each converted pixel obtained by the addition value calculation means 130 is stored in the memory device 200. The maximum value of the addition value in the small area is added to the reference pixel addition value 240, and an identifier (intermediate area) indicating whether the small area classified by the area classification unit 120 is a halftone area or a binary area. A key binary identifier) is also added as an attribute.
[0032]
[Conversion pixel value calculation means 140]
As shown in FIG. 5, the conversion pixel value calculation means 140 is composed of a conversion pixel value calculation means 141 for a halftone area and a conversion pixel value calculation means 142 for a binary area. The converted pixel value calculating means 141 for the halftone area receives the reference pixel added value 240 and refers to the halftone / binary identifier added as an attribute, and is classified as a halftone area by the area classifying means 120. Sometimes processing is performed. The converted pixel value calculating means 142 for the binary area is processed when the area classifying means 120 classifies it as a binary area.
[0033]
Here, the original imageHalftoneLet A be the maximum value of the gradation, and B be the maximum value of the gradation of the halftone of the converted image. The white value is 0, and the maximum black density is A or B. For example, in the case of 256 gradations, possible gradation values are 0 to 255, where 0 indicates white and 255 indicates black. Also,134Let M be the maximum value in the small area calculated in.
[0034]
Here, when the corresponding small area is classified into the halftone area and the area classification unit 120, the value of the conversion pixel Qij is calculated by the conversion pixel value calculation unit 141 of the halftone area.
[0035]
(Equation 3)

[0036]
Is calculated as Conversely, when the corresponding small area is classified as a binary area, the value of the conversion pixel Qij is calculated by the conversion pixel value calculation means 142 of the binary area.
[0037]
(Equation 4)

[0038]
Is calculated as Note that when the original image is a binary image, the calculation can be performed by setting A = 1. As described above, in the case of the halftone area, the added value is divided by the number of original image gradations and the number of reference pixels, while in the case of the binary area, the added value is divided by the maximum value M in the small area to obtain the value of the converted pixel. . At this time, M ≦ A × n × n. This M is the maximum value in the small area, and holds the value with the highest density. The value of the converted pixel in the halftone area is obtained by dividing rij by (A × n × n), that is, (the number of original image gradations × the number of reference pixels). In the binary area, rij is divided by M to obtain the converted pixel value. This has the effect of suppressing the value from becoming thin. This is especially true in the binary domain where the conversion ratio isGet smallerAccordingly, the number of reference pixels to be added (n × n) increases, and the added value is leveled, so that a phenomenon occurs in which the value of the strongest density becomes lighter. Therefore, in the binary area, the maximum value M of the addition value in the small area is obtained, and the maximum black density is replaced with (the number of original image gradations × the number of reference pixels). However, there is a great effect that thin lines of the original image can be prevented from becoming thin.
[0039]
The output of the conversion pixel value calculation means 140 is stored in the memory device 250 as the conversion image 250. The converted image 250 can be output to the image output device 400 as it is.
[0040]
[Converted image correction means 150]
As shown in FIG. 6, the converted image correcting unit 150 includes a halftone output value determining unit 151 and a binarizing unit 152.
The halftone output value determination unit 151 is a unit that takes consistency when the number of tones that can be expressed by the output device does not match the number of tones that the converted image has. When the converted image is created as a converted image capable of expressing up to 512 gradations, values of 0 to 511 can be expressed with B = 511. The number of tones to be expressed is finally changed according to the output device once the converted image is created. For example, in the case of an output device that can express only four gradations, 256 gradations are assigned to a limited number of halftone values of four gradations, and a value to be actually output is determined. Conversely, when the output device can represent 256 gradations and the image has only 4 gradations, the value expressed and distributed equally every 64 gradations without white or black bias is determined.
[0041]
Note that the halftone output value determining means 151 may not be an essential means for the present invention. For example, there are output devices that can automatically assign the number of tones to be output automatically and the four tones when a halftone image of 256 tones is directly input to an output device capable of four tones. When outputting to such a device, the present invention does not require the halftone output value determining means 151. This does not limit the scope of the invention.
[0042]
The binarizing unit 152 is a unit that sets a threshold value corresponding to each of the halftone area and the binary area, binarizes the converted image of the halftone expression, and outputs a binary image. This process is as follows. As an example, the halftone image subjected to the scale-down conversion is B = 255, that is, an image of 256 gradations is created. The value of s1 for the halftone region and the value of s2 for the binary region are determined in advance for the halftone region and the binary region. For example, s1 = 128 and s2 = 64. When s1 and s2 are stored, the binarization process of s1 is performed when the region is divided into the halftone region by the region dividing unit 120, and when the region is classified into the binary region by the region dividing unit 120, the binary value of s2 is stored. Is implemented. Thereby, binarization of the halftone image which has been reduced and converted is performed.Can be realizedIt is. Furthermore, by separately setting the threshold values for the halftone area and the binary area, the optimum threshold value for the halftone area and the optimum threshold value for the binary area can be set, and the best converted image quality can be set for each. Can be created.
[0043]
As described above, the binarization unit 152 can binarize the halftone image according to the present embodiment to generate a binary image. As for the binarization, there are a binarization unit for one conversion pixel and a binarization method of collecting a plurality of conversion pixels, and both can be applied to the binarization unit 152. It is. Binarizing only one conversion pixel has the advantages of simple processing and small hardware. The method of collecting a plurality of converted pixels and binarizing it requires complicated processing and apparatus, but a conventional method that does not impair the gradation characteristics when binarizing a halftone area such as a photograph can be applied. This has the advantage that binarization can be performed while suppressing quality deterioration.
[0044]
In the description of the above embodiment, when determining the value of the conversion pixel divided into the binary area and the halftone area, the halftone area uses a method of dividing the added value by the number of reference pixels and the number of original image gradations. In the value area, the method of dividing the added value rij by the maximum value M in the small area is shown. On the other hand, another embodiment of the present invention which realizes the coefficient α1 in the halftone area and the coefficient α2 in the binary area in advance by using α1 and α2 for each area is described below. Shown in Here, when the small area is classified by the halftone area and the area dividing means 120, the converted pixel value calculating means 141 of the halftone area sets the value of the converted pixel Qij as
[0045]
(Equation 5)

[0046]
Is calculated and obtained. When the small area is classified as a binary area by the area classifying means 120, the converted pixel value calculating means 142 for the binary area
[0047]
(Equation 6)

[0048]
Is calculated and obtained. When the original image is a binary image, A = 1.
[0049]
As described above, when the small area is the halftone area, the value of the conversion pixel is calculated by multiplying by the coefficient α1 and when the small area is the binary area, by multiplying by α2. By using these two coefficients, a converted image can be created without using the maximum value calculation means 134, that is, the maximum value M.
[0050]
This is because when the conversion ratio is particularly small, the characters and thin lines in the binary area become thin, and the quality is greatly deteriorated. For this reason, for the binary area, the coefficient α2 is increased, the density of the converted pixel is emphasized to the black density, and the optimum quality of the converted image generated thereby is generated. On the contrary, in the halftone area, α1 is set to a value close to 1 because there is no such phenomenon of thinning. That is, it is a major feature of the present invention that α1 is set in the halftone area and α2 is set in the binary area, and α1 ≦ α2. In particular, only the converted pixels in the binary area are emphasized in black density. Here, as specific examples of the coefficients α1 and α2, for example, α1 = 1 and α2 = 3 are determined. Thus, the coefficients α1 and α2 are defined in advance. Note that the values of α1 and α2 are merely examples, and are not fixed to these values, but can be set to optimal values depending on the type of output device.
[0051]
As described above, the area of the original image is divided into a binary area and a halftone area, the coefficient α2 is determined so as to emphasize the grayscale value of the converted pixel in the binary area, and the grayscale characteristic of the halftone area is determined. .Alpha.1 is determined so as not to deteriorate the image quality. Since the means for determining the conversion pixel for each area is provided, it is possible to prevent the image quality from deteriorating, thereby having an extremely large effect.
[0052]
As described above, it is a feature of the present invention that the original image is divided into a halftone region and a binary region, and each unit has a unit for calculating the value of the converted pixel.
[0053]
Further, a feature of the present invention is to have a means for dividing the added value of the reference pixel by the number of original image gradations and the number of reference pixels in the halftone area and dividing the added value by the maximum value M in the small area in the binary area. .
[0054]
In addition, without obtaining the maximum value M of the small area, the coefficient α1 is determined in advance for the halftone area and the coefficient α2 is determined for the binary area, and the sum of the added value and each coefficient is used to calculate the conversion pixel. It is a feature of the invention to implement another means of determining the value.
[0055]
Note that, in the present invention, the case where the conversion ratio in the horizontal direction and the vertical direction is the same has been described, but the case where the conversion ratio in the horizontal direction and the vertical direction are different, and one of them is also applicable to the enlargement conversion. .
[0056]
In the present invention, the case where both the original image and the converted image are black and white halftone images or black and white binary images has been described. However, the present invention is not limited to black and white images. In addition to a black and white image, the present invention can be applied to, for example, an image expressed in two colors such as red characters on a white background. That is, if the presence / absence of information is expressed by, for example, the maximum value 255 and the minimum value 0 of the halftone image, 255 can be reduced to a halftone image according to the present technology using red as a base and 0 as a white base. The halftone image reduced and converted in this way is represented by two colors of white and red by expressing the color having the maximum gradation as red and the lowest gradation as a white background and expressing the intermediate density with the red gradation component. Is reduced to a halftone image. Further, the result of the binarization processing is equivalent to coloring the place where the information is present in red and the place where the information is not present on the white background.
[0057]
As described above, one embodiment of the present invention has been described. However, the respective means 110 to 150 in FIG. 1 and the specific realizing means shown in FIGS. 2 to 6 may be realized using a computer program. A program in which the functions, procedures, and algorithms are described so as to be executable by a computer can be provided by being recorded on a floppy disk, a memory card, a CD-ROM, or the like.
[0058]
【The invention's effect】
According to the present invention, when outputting to an output device capable of expressing a halftone image, the original image is classified into a halftone region and a binary region, and the density value of the converted pixel is calculated for each region. This makes it possible to create and output a halftone image of optimal quality.
In addition, in the halftone area, the sum of the reference pixels is calculated as the number of reference pixels.Original imageIn the binary area, the divided value is divided by the maximum value of the small area, and the value of the converted pixel is obtained, whereby the optimum value of the converted pixel can be calculated.
Also, a coefficient α1 is provided in the halftone area, and a coefficient α2 is provided in the binary area. By multiplying the addition value by a coefficient, the optimum value of the conversion pixel for each area can be obtained without using the maximum value M. Is required.
As described above, the best converted image with little image quality deterioration can be created in both the halftone region and the binary region. In particular, when the conversion ratio is reduced, or when the output device has a higher resolution, the converted image is expressed thinly in the binary region, and the converted image quality is significantly deteriorated. It can be overcome by the invention.
[Brief description of the drawings]
FIG. 1 is a block diagram of one embodiment of the present invention.
FIG. 2 is a diagram illustrating a configuration example of an original image type detection unit.
FIG. 3 is a diagram illustrating a configuration example of a region classification unit.
FIG. 4 is a diagram illustrating a configuration example of a reference pixel addition value calculation unit.
FIG. 5 is a diagram illustrating a configuration example of a conversion pixel value calculation unit.
FIG. 6 is a diagram illustrating a configuration example of a converted image correction unit.
FIG. 7 is a diagram showing an example in which an original image is divided into small areas and there are a binary area and a halftone area.
FIG. 8 is a diagram illustrating an example of a technique for realizing area classification into a halftone area and a binary area.
FIG. 9 is a diagram illustrating a specific example of a positional arrangement relationship between original pixels and converted pixels.
[Explanation of symbols]
110 Original image type detecting means
120 area classification means
130 Reference pixel addition value calculation means
140 Conversion pixel value calculation means
150 Converted image correction means

Claims (5)

In an image reduction converter for outputting an original image reduced by an arbitrary conversion ratio,
Means for dividing the original image into a plurality of small areas;
Means for classifying the divided small areas into halftone areas or binary areas;
For each of the small areas, a pixel of the converted image is defined as a conversion pixel, and when the conversion pixels are arranged so as to be superimposed on respective corresponding positions on the original image, the number of pixels corresponding to the conversion ratio around each conversion pixel is determined. Means for obtaining an added value of the values of the reference pixels, using the pixels of the original image as reference pixels, and obtaining a maximum value of the added value of the small area,
In the case of the small area classified as the halftone area, the added value is divided by a value obtained by multiplying the number by the number of original image gradations to obtain a value of a converted pixel, and the small area classified into the binary area is obtained. In the case of a region, means for dividing the added value by the maximum value to obtain a value of a converted pixel,
An image reduction conversion device comprising: In an image reduction converter for outputting an original image reduced by an arbitrary conversion ratio,
Means for dividing the original image into a plurality of small areas;
Means for classifying the divided small areas into halftone areas or binary areas;
For each of the small areas, a pixel of the converted image is defined as a conversion pixel, and when the conversion pixels are arranged so as to be superimposed on respective corresponding positions on the original image, the number of pixels corresponding to the conversion ratio around each conversion pixel is determined. Means for obtaining an added value of the values of the reference pixels, using the pixels of the original image as reference pixels, and obtaining a maximum value of the added value of the small area,
In the case of the small area classified as the halftone area, a value obtained by multiplying the added value by a predetermined coefficient α1 is divided by the number to obtain a value of the conversion pixel, and the small area classified into the binary area is obtained. In the case of a region, a value obtained by multiplying each of the added values by a predetermined coefficient α2 (α1 ≦ α2) is divided by the number to obtain a converted pixel value. ,
An image reduction conversion device comprising: In a reduction conversion method of an image in which an original image is reduced and output at an arbitrary conversion ratio,
Processing of dividing the original image into a plurality of small areas;
A process of classifying the divided small region into a halftone region or a binary region;
For each of the small areas, a pixel of the converted image is defined as a conversion pixel, and when the conversion pixels are arranged so as to be superimposed on respective corresponding positions on the original image, the number of pixels corresponding to the conversion ratio around each conversion pixel is determined. Assuming that the pixels of the original image are the reference pixels, the sum of the values of the reference pixels is obtained, and the maximum value of the sum of the small areas is obtained,
In the case of the small area classified as the halftone area, each of the added values is divided by a value obtained by multiplying the number by the number of original image gradations to obtain a value of the converted pixel, and the small value classified into the binary area is obtained. In the case of a region, a process of dividing the added value by the maximum value to obtain a value of a converted pixel,
An image reduction conversion method comprising: In a reduction conversion method of an image in which an original image is reduced and output at an arbitrary conversion ratio,
Processing of dividing the original image into a plurality of small areas;
A process of classifying the divided small region into a halftone region or a binary region;
For each of the small areas, a pixel of the converted image is defined as a conversion pixel, and when the conversion pixels are arranged so as to be superimposed on respective corresponding positions on the original image, the number of pixels corresponding to the conversion ratio around each conversion pixel is determined. Assuming that pixels of the original image are reference pixels, a process of obtaining the maximum value of the addition value of the small area, while calculating the sum of the values of the reference pixels,
In the case of the small area classified as the halftone area, a value obtained by multiplying the added value by a predetermined coefficient α1 is divided by the number to obtain a value of the conversion pixel, and the small area classified into the binary area is obtained. In the case of a region, a value obtained by multiplying the added value by a predetermined coefficient α2 (α1 ≦ α2) is divided by the number to obtain a converted pixel value;
An image reduction conversion method comprising: A computer-readable recording medium which records a program for executing the image reduction conversion method according to claim 3 on a computer.

Images