201141243 六、發明說明： 【發明所屬之技術領域】 本發明係有關於一種梳型濾波裝置，特別是有關於一 種可根據一影像信號的變化，適當地處理該影像信號的梳 型濾波裝置。 & 【先前技術】 梳型濾波器（comb filter)在傳統電視機（類比電視）上是 一個相當重要的電路。梳型濾波器最主要的工作原理是， 利用彩色訊號與黑白訊號之間相位的不同，分離出彩色訊 號與黑白訊號。分離後的彩色訊號與黑白訊號分別送至各 自的解調器。 如果彩色訊號與黑白訊號分離的乾淨，則解調後的，奪 ^與彩色訊號就會完全的分離，不會干擾彼此。如此就能 ，原出完美的原像。相反地’若彩色訊號與黑自訊號無法 完整地分離時，則解調後的黑白與彩色訊號就會互相干 擾’使晝面（圖框）出現拍差(beat)以及額外的彩色條紋，進 而造成觀賞的不適。 【發明内容】 本發明提供—種梳㈣波裝置，接收一連續圖場訊 2分離—目前圖場的-目前影像成份的亮度及彩度，並 it處理單元以及一控制單元。處理單元藉由對目前影 卜二進彳2D梳型;慮波處理以及—犯梳型遽波處理之 成二者，分離目前影像成份的亮度及彩度。當目前影像 成伤進行扣梳型遽波處理後，可產生一第一處理信號。 201141243 當目前影像成份進行3D梳型濾波處理後，可產生第二處 理信號。控制單元包括，一第一計算模組、一第二計算模 組以及一判斷模組。一第一計算模組對目前圖場的目前影 像成份、複數第一影像成份、一第一先前圖場的一第一先 前影像成份及複數第二影像成份進行一絕對值差和計算， 用以產生一第一計算結果。第二計算模組對目前影像成 份、該等第一影像成份、一第二先前圖場的一第二先前影 像成份及複數第三影像成份，進行絕對值差和計算，用以 • 產生一第二計算結果。一判斷模組根據第一及第二計算結 果，產生一第一判斷結果。當第一判斷結果大於一第一預 設值時，判斷模組令處理單元輸出第一處理信號。當第一 判斷結果小於第一預設值時，判斷模組令處理單元輸出第 二處理信號。 為讓本發明之特徵和優點能更明顯易懂，下文特舉出 較佳實施例，並配合所附圖式，作詳細說明如下： 【實施方式】 * 第1圖為本發明之梳型濾波裝置之一可能實施例。梳 型濾波裝置100用以處理影像信號VIN1。第6圖為影像信 號V1N1之示意圖。如第6圖所示，影像信號VIN1係為具有 連續圖框(frame)PF 1、PF2、CF的影像信號（即連續圖場(field) F、F0〜F4)。稍後將詳細說明第6圖。在本實施例中，梳型 濾波裝置100包括處理單元110以及控制單元130。 在一可能實施例中，梳型濾波裝置100可整合在一類 比數位轉換器（analog-to-digital converter ; ADC)中，但並 201141243 非用以限制本發明。在其它實施例中，梳型濾波裝置100 可應用在影像解碼器（video decoder)之中。 另外’本發明並不限定影像信號VIN1的種類。在一可 能實施例中，影像信號Vwi係為NTSC(National Television System Committee)規格的影像信號（即每秒60張顯示晝 面）’但並非用以限制本發明。在其它實施例中，影像信號 V】ni係為PAL(Phase Alternating Line)規格，或是其它規格 的影像信號。 以NTSC規格為例，影像信號VlN1中每一圖場均具有 許多影像成份，每一影像成份被供予一對應的像素單元。 一顯示器（未繪出）具有複數掃描線，每一條掃描線耦接複 數個像素單元。該等像素單元根據影像信號VlN1的影像成 份，呈現相對應的晝面。奇圖場(odd field)，如第6圖的圖 場F、FI、F3 ’具有複數影像成份，該等影像成份係供予 複數奇數掃描線。同樣地，偶圖場(even field)，如第6圖 的圖場F0、F2、F4，亦具有複數影像成份，該等影像成 份係供予複數偶數掃描線。一般來說，顯示晝面之前會先 對圖場(如F、F0-F4)做梳型濾波處理，先將圖場中的各影 像成份的亮度和色度作分離，之後再進行解交錯 (de-interlace)處理，最後才於顯示器上顯示出對應晝面。 為方便說明’以下内容僅以影像信號Vwi的偶圖場（如 第6圖的F0、F2、F4)的影像成份F〇_yc、F2_yc及F4__yc 為例’說明梳型濾波裝置100如何處理影像信號VIN】的影 像成份F0一yc、F2__yc及F4_yc，其中影像成份F〇_yc、F2一yc 及F4一yc均係欲提供予一目標像素pix的影像成份。上述 201141243 影像成份F0_yc、F2_yc及F4_yc在提供予目標像素Pix之 前’需經過2D或3D濾波處理以分離亮度及彩度。目標像 素Pix再根據處理後的結果，呈現相對應的亮度及色度。 在本實施例中，請參照第6圖，影像成份F0_yc為一 第一圖框CF的偶圖場F0的影像成份；影像成份F2_yc為 —第二圖框PF2的偶圖場F2的影像成份；影像成份F4__yc 為一第三圖框PF1的偶圖場F4的影像成份，其中第一至第 三圖框PF1、PF2、CF為連續圖框。在本實施例中，第三 圖框P F1為最早接收到的圖框，然後才接收到第二圖框p F 2 為’最後接收到第一圖框CF。因此，第三圖框PF1稱為第 —先前圖框，第二圖框PF2稱為第二先前圖框，第一圖框 CF稱為目前圖框。 偶圖場F4的影像成份F4_yc最早被提供，接著提供偶 圖場F2的影像成份F2_yc，最後提供偶圖場F0的影像成 份F0_yc。因此，影像成份F4_yc可稱為第一先前影像成 份，影像成份F2_yc可稱為第二先前影像成份，而影像成 份F0_yc可稱為目前影像成份。 影像成份F0_yc、F2_yc及F4_yc係為圖場上’欲提供 予同一像素（如圖場座標（1〇，1〇)位置的點像素）的影像信 號。另外，由於影像信號Vmi具NTSC規格’故影像成份 F〇_yc、F2_yc 及 F4_yc 同時具有亮度成份(luminance ; Y) 及色度成份(chrominance ; C) ’且影像成份F0—yc、F2_yc 及F4_yc的色度成份像位差為180度。 在第1圖中’處理單元13〇對目前圖框CF的目前影像 成份(如F0_yc)進行一 2D梳槊濾波處理以及一 3D梳型濾 201141243 波處理之至少一者。當目前影像成份進行完2D梳型濾波 處理後，可產生處理信號SP1 2D。當目前影像成份進行完 3D梳型濾波處理後，可產生處理信號Sp2 3D。在一可能實 %例中’處理彳§號Spi_2D或Sp2 3D係提供予一像素（pixel) ’ 使得該像素呈現相對應的亮度及色度。 在本實施例中，處理單元no包括，2D梳型濾波器（2D combfilter)lll 以及 30梳型濾波器（3Dcombfilter)113。當 2D梳型濾波器111被觸發時，便對目前影像成份進行2D 梳型濾波處理’用以產生處理信號SP1_2D。當3D梳型濾波 器113被觸發時’便對目前影像成份進行3D梳型濾波處 理’用以產生處理信號SP2 3D。在一實施例中，一目標像 素（如Pix)根據處理信號兮〜瓜或SP2 3D，而呈現相對應亮 度及色度。 如圖所示，控制單元13〇包括，計算模組131、133以 及判斷模組135。計算模組131對目前影像成份F0_yc、複 數第一影像成份(如第6圖範圍600内，除了 F0_yc以外的 所有影像成份）、第一先前影像成份F4_yc以及複數第二影 像成份(如第6圖範圍604内，除了 F4_yc以外的所有影像 成份）進行一絕對值差和（sum of absolute differences ;以下 簡稱SAD)計算，用以產生計算結果Sci。在本實施例中， 根據計鼻結果SC!，便可得知目前影像成份J?〇_yc相對於先 前影像成份F4_yC而言，是否為靜止影像成份。 請配合第6圖，計算模組131係對欲提供至範圍6〇〇 及604内的像素的相對應影像成份進行SAD計算。在本實 施例中’範圍600及604各自包含目標像素Pix相鄰的8 201141243 個《。本發明並不限制範圍6。。 其它貫施例中，計算模紐i 31 及6〇4的大小。在 SAD計算。 可對更多的影像成份進行 為了提高判斷的準確性，本 計算模組133。藉由計算模組13/， j控制單元130具有 目標像素Pix的影像成份、F〇 ye j便可更加確認欲傳送至 實施例中，為了節省成本，靜止成份。在其它 在本實施财，計⑶。 複數第-影像缝(如第6圖_ _ ;，=成y 的所有影像餘）、第二先前影像成份F2 外 影像成份(如第ό圖範圍602内，除 數第一 像成份)進行SAD計算，用以產生; ▲丄外^ 王0丁开、、、口果SC2。如圖所示， 计异模組133包括，内部2D梳型滤波器132、134 算器136。 内部2D梳型濾、波器132處理欲提供至範圍6〇2内的像 素的影像成份，並產生複數内部信號F2一y。在本實施例中， 内部信號F2一y為欲提供至範圍602内的像素的影像成份的 亮度成份。 内部2D梳型〉慮波器134處理欲提供至範圍6〇〇内的像 素的影像成份’並產生複數内部信號F0_y。在本實施例中， 内部信號F0_y為欲提供至範圍600内的像素的影像成份的 亮度成份。計算器136對内部信號F2_y及F〇進行sad 計算，用以產生計算結果SC2。 判斷模組135根據計算結果SC2 ’可更加確認欲提供至 目標像素Pix的影像成份F0_yc是否為靜止影像成份。在 201141243 本實施例中，判斷模組135係同時根據計算結果Sci及BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a comb filter device, and more particularly to a comb filter device that can appropriately process a video signal according to a change in an image signal. & [Prior Art] The comb filter is a very important circuit on a conventional television (class analog TV). The main working principle of the comb filter is to separate the color signal and the black and white signal by using the phase difference between the color signal and the black and white signal. The separated color signals and black and white signals are sent to the respective demodulator. If the color signal is separated from the black and white signal, the demodulated and fully separated color signals will not interfere with each other. This way, the original original image is original. Conversely, 'if the color signal and the black signal cannot be completely separated, the demodulated black and white and color signals will interfere with each other', causing the beat (bezel) and extra color stripes to appear. Causes discomfort in viewing. SUMMARY OF THE INVENTION The present invention provides a comb (four) wave device that receives a continuous picture field 2 separation - current picture field - brightness and chroma of the current image component, and it processing unit and a control unit. The processing unit separates the brightness and chroma of the current image components by the two-in-one 2D comb type; the wave processing and the comb-type chopping processing. When the current image is injured and subjected to a comb-type chopping process, a first processed signal can be generated. 201141243 When the current image component is subjected to 3D comb filtering, a second processing signal can be generated. The control unit includes a first computing module, a second computing module, and a determining module. a first computing module performs an absolute difference between the current image component of the current field, the plurality of first image components, a first previous image component of the first previous field, and the plurality of second image components, A first calculation result is generated. The second computing module performs absolute value difference calculation on the current image component, the first image component, a second previous image component of the second previous field, and the plurality of third image components to generate a first Second, calculate the result. A judging module generates a first judgment result based on the first and second calculation results. When the first determination result is greater than a first preset value, the determination module causes the processing unit to output the first processing signal. When the first judgment result is less than the first preset value, the judging module causes the processing unit to output the second processing signal. In order to make the features and advantages of the present invention more comprehensible, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 1 is a comb filter of the present invention. One of the possible embodiments of the device. The comb filter device 100 is for processing the image signal VIN1. Figure 6 is a schematic diagram of the image signal V1N1. As shown in Fig. 6, the video signal VIN1 is a video signal having continuous frames PF 1, PF2, CF (i.e., continuous fields F, F0 to F4). Fig. 6 will be described in detail later. In the present embodiment, the comb filter device 100 includes a processing unit 110 and a control unit 130. In a possible embodiment, the comb filter device 100 can be integrated into an analog-to-digital converter (ADC), and 201141243 is not intended to limit the invention. In other embodiments, the comb filter device 100 can be used in a video decoder. Further, the present invention does not limit the type of the video signal VIN1. In a possible embodiment, the video signal Vwi is an NTSC (National Television System Committee) video signal (i.e., 60 display screens per second), but is not intended to limit the present invention. In other embodiments, the image signal V]ni is a PAL (Phase Alternating Line) specification or an image signal of other specifications. Taking the NTSC specification as an example, each field in the image signal VlN1 has a plurality of image components, and each image component is supplied to a corresponding pixel unit. A display (not shown) has a plurality of scan lines, each of which is coupled to a plurality of pixel units. The pixel units present corresponding pupil planes according to the image component of the image signal V1N1. The odd field, such as the fields F, FI, F3' of Figure 6, has a plurality of image components that are supplied to a plurality of odd scan lines. Similarly, the even field, such as field F0, F2, and F4 of Figure 6, also has a plurality of image components that are supplied to a plurality of even scan lines. In general, before the display of the surface, the comb field filtering process (such as F, F0-F4) is performed first, and the brightness and chromaticity of each image component in the field are separated first, followed by deinterlacing ( De-interlace), and finally the corresponding face is displayed on the display. For convenience of explanation, the following description only uses the image components F〇_yc, F2_yc, and F4__yc of the image field Vwi (for example, F0, F2, and F4 of FIG. 6) as an example to describe how the comb filter device 100 processes images. The image components of the signal VIN are F0-yc, F2__yc, and F4_yc, wherein the image components F〇_yc, F2-yc, and F4-yc are image components to be supplied to a target pixel pix. The above-mentioned 201141243 image components F0_yc, F2_yc, and F4_yc are subjected to 2D or 3D filtering processing to separate luminance and chroma before being supplied to the target pixel Pix. The target pixel Pix then displays the corresponding brightness and chromaticity according to the processed result. In this embodiment, referring to FIG. 6, the image component F0_yc is an image component of the even field F0 of the first frame CF; the image component F2_yc is an image component of the even field F2 of the second frame PF2; The image component F4__yc is an image component of the even field F4 of the third frame PF1, wherein the first to third frames PF1, PF2, and CF are continuous frames. In the present embodiment, the third frame P F1 is the earliest received frame, and then the second frame p F 2 is received as the last received first frame CF. Therefore, the third frame PF1 is referred to as a first previous frame, and the second frame PF2 is referred to as a second previous frame, and the first frame CF is referred to as a current frame. The image component F4_yc of the even field F4 is provided first, then the image component F2_yc of the even field F2 is provided, and finally the image component F0_yc of the even field F0 is provided. Therefore, the image component F4_yc may be referred to as a first previous image component, the image component F2_yc may be referred to as a second previous image component, and the image component F0_yc may be referred to as a current image component. The image components F0_yc, F2_yc, and F4_yc are image signals on the field to be supplied to the same pixel (the dot pixel at the position of the field coordinates (1〇, 1〇)). In addition, since the image signal Vmi has the NTSC specification, the image components F〇_yc, F2_yc, and F4_yc have both luminance components (luminance; Y) and chrominance components (C)' and image components F0-yc, F2_yc, and F4_yc. The chroma component has a position difference of 180 degrees. In Fig. 1, the processing unit 13 进行 performs at least one of a 2D comb filtering process and a 3D comb filtering 201141243 wave processing on the current image component (e.g., F0_yc) of the current frame CF. When the current image component is subjected to 2D comb filtering processing, the processing signal SP1 2D can be generated. When the current image component is subjected to 3D comb filtering processing, the processing signal Sp2 3D can be generated. In a possible example, the processing 彳§ Spi_2D or Sp2 3D is provided to a pixel (pixel) such that the pixel exhibits corresponding brightness and chrominance. In the present embodiment, the processing unit no includes a 2D comb filter 111 and a 3D comb filter 113. When the 2D comb filter 111 is triggered, the current image component is subjected to 2D comb filtering processing to generate the processed signal SP1_2D. When the 3D comb filter 113 is triggered, the 3D comb filter processing is performed on the current image component to generate the processed signal SP2 3D. In one embodiment, a target pixel (e.g., Pix) exhibits corresponding brightness and chromaticity based on the processed signal 瓜~ melon or SP2 3D. As shown, control unit 13A includes computing modules 131, 133 and decision module 135. The calculation module 131 has a current image component F0_yc, a plurality of first image components (such as all image components except F0_yc in the range 600 of FIG. 6), a first previous image component F4_yc, and a plurality of second image components (such as FIG. 6). In the range 604, all image components except F4_yc are subjected to a sum of absolute differences (SAD) calculation to generate a calculation result Sci. In this embodiment, based on the nasal result SC!, it can be known whether the current image component J?〇_yc is a still image component with respect to the previous image component F4_yC. In conjunction with FIG. 6, the calculation module 131 performs SAD calculation on the corresponding image components of the pixels to be provided to the ranges 6〇〇 and 604. In the present embodiment, the ranges 600 and 604 each include 8 201141243 adjacent to the target pixel Pix. The invention is not limited to the scope 6. . In other embodiments, the sizes of the moduli i 31 and 6 〇 4 are calculated. Calculated in SAD. More image components can be used to improve the accuracy of the judgment, the calculation module 133. By the calculation module 13/, j control unit 130 having the image component of the target pixel Pix, F 〇 ye j can be further confirmed to be transmitted to the embodiment, in order to save cost, the stationary component. In other implementations in this implementation, (3). The complex first-image seam (such as Figure 6 _ _ ;, = all images of y), the second previous image component F2 external image components (such as the first image component in the range 602 of the first image) for SAD Calculation, used to produce; ▲ 丄 外 ^ Wang 0 Ding Kai,,, fruit fruit SC2. As shown, the add-on module 133 includes an internal 2D comb filter 132, 134 calculator 136. The internal 2D comb filter, waver 132 processes the image components to be supplied to the pixels in the range of 6 〇 2 and generates a complex internal signal F2 - y. In the present embodiment, the internal signal F2_y is the luminance component of the image component to be supplied to the pixels in the range 602. The internal 2D comb type filter 134 processes the image component 'to be supplied to the pixels within the range of 6 ’' and generates a complex internal signal F0_y. In the present embodiment, the internal signal F0_y is the luminance component of the image component to be supplied to the pixels within the range 600. The calculator 136 performs a sad calculation on the internal signals F2_y and F〇 to generate a calculation result SC2. The judging module 135 can further confirm whether or not the image component F0_yc to be supplied to the target pixel Pix is a still image component based on the calculation result SC2'. In the embodiment of 201141243, the determination module 135 is based on the calculation result Sci and
Sc2 ’產生判斷結果SD1。本發明並不限定判斷模組135產 生判斷結果SD1的方法。 在一可能實施例中，判斷模組135平均計算結果So 及SC2用以產生判斷結果。在其它實施例中，判斷結 果SD1係為計算結果Sa及之最大者。在本實施例中， 判斷模組135具有一混合器（Μίχ)137，用以混合計算結果 So ’而產生判斷結果sDi。 當判斷結果sD1大於一第一預設值時，判斷模組135 令處理單元110輸出處理信號Spi_2D。在本實施例中，當 判斷結果SD1大於第—預設值時，表示欲提供至目標像素 Pix的目前影像成份F〇__yC相對於先前影像成份(如F2_yc 或F4—yC)而言’係對應一移動後的影像成份，故判斷模組 135觸發2D梳型濾波器111，用以對欲提供至目標像素pix 目前影像成份F0_yc進行2D梳型濾波處理，然後再將渡 波處理後的結果傳送至目標像素Pix。 當判斷結果SD1小於第一預設值時’判斷模組135令處 理單元110輸出處理信號SP2 3D。在本實施例中，當判斷 結果SD]小於第一預設值時’表示欲提供至目標像素Pix 的目前影像成份抑-^相對於先前影像成份F2-yc4F4_yc 而言，係對應到一靜止影像成份，故判斷模組13 5觸發3d 梳型濾波器113,用以對欲提供至目標像素pix所接收到的 目前影像成份F〇__yc進行3D梳型遽波處理。3D梳型濾波 處理後的結果亦將傳送至目標像素Pix。 第2圖為本發明之梳型濾波裝置之另一可能實施例。 201141243 如圖所示’梳型濾波裝置200包括’處理單元2i〇以及控 制單元230。由於控制單元230與第1圖的控制單元^3〇 相似，故不再贅述。在本實施例中，處理單元21〇包括， 2D梳型濾波器21 l、3D梳型濾波器213以及開關模組215。Sc2' produces a judgment result SD1. The present invention does not limit the method by which the determination module 135 generates the determination result SD1. In a possible embodiment, the determining module 135 averages the calculated results So and SC2 to generate a determination result. In other embodiments, the judgment result SD1 is the calculation result Sa and the largest one. In the present embodiment, the judging module 135 has a mixer 137 for mixing the calculation result So ′ to generate the judgment result sDi. When the determination result sD1 is greater than a first preset value, the determination module 135 causes the processing unit 110 to output the processing signal Spi_2D. In this embodiment, when the determination result SD1 is greater than the first preset value, it indicates that the current image component F〇__yC to be supplied to the target pixel Pix is relative to the previous image component (such as F2_yc or F4-yC). Corresponding to a moving image component, the determining module 135 triggers the 2D comb filter 111 for performing 2D comb filtering processing on the current image component F0_yc to be supplied to the target pixel pix, and then transmitting the result of the wave processing. To the target pixel Pix. When the judgment result SD1 is smaller than the first preset value, the judgment module 135 causes the processing unit 110 to output the processing signal SP2 3D. In this embodiment, when the determination result SD] is smaller than the first preset value, 'the current image component to be supplied to the target pixel Pix is compared with the previous image component F2-yc4F4_yc, corresponding to a still image. In the component, the judging module 13 5 triggers the 3d comb filter 113 for performing 3D comb chopping processing on the current image component F〇__yc to be supplied to the target pixel pix. The result of the 3D comb filter processing will also be transmitted to the target pixel Pix. Figure 2 is another possible embodiment of the comb filter device of the present invention. 201141243 The comb filter device 200 includes a processing unit 2i and a control unit 230 as shown. Since the control unit 230 is similar to the control unit 3 of FIG. 1, it will not be described again. In the present embodiment, the processing unit 21A includes a 2D comb filter 21 l, a 3D comb filter 213, and a switch module 215.
2D梳型濾波器211對欲傳送至目標像素pix的目前影 像成份F0_yc進行2D梳型濾波處理，用以產生處理信號 Sp〗_2〇。3D梳型滤波态213對欲傳送至目標像素pix的目 前影像成份F0_yc進行3D梳型濾波處理，用以產生處理 信號Spud。開關模組215接收處理信號％】①及心2扣， 並根據判斷模組235的輸出信號，決定輸出處理信號 或SP2_3D予目標像素pix。 一 在本實施例中，判斷模組235根據計算模組23丨及233 所產生的計算結果Sg]及，產生判斷結果^。當判斷 =SD1大於一第一預設值時，判斷模址故控制開關模 j⑺’使其輸出處理信號Sp]_2D。當判斷結果s⑴小於第 : = 斷模組235控制開關模組-，使其輸出 栌型2 =例中，在控制單元230開始運作的同時，犯 1及3D梳型遽波器213便開始對欲傳送至 不素Ριχ的目前影像成份F〇 開關模組215她㈣斷模組23;的輸; 即輸==處理信〜或、二=立 &制早兀細多了計算模組37G及。由於控制單元30的0 201141243 的計算模組310、330以及判斷模組350與第1圖的計算模 組131、133以及判斷模組135相似’故不再贅述。 在第3圖中’計算模組370對欲傳送至目標像素Pix 的目前影像成份F〇_yc進行〆邊緣判斷(edge detection)處 理’用以產生計算結果SC3。在本實施例中，計算模組370 係對欲傳送至目標像素pix的目前影像成份F〇_yC的亮度 成份F0__y進行處理（例如，高頻濾波處理），用以產生一邊 緣資訊。 計算模組390根據欲傳送至目標像素Pix的目前影像籲 成份F0一yc及第二先前影像成份F2_yc ’進行邊緣判斷處 理’用以產生計算結果SC4。藉由計算模組39〇，便可更加 確έ忍欲傳送至目標像素ρ[χ的目前影像成份是否為 —邊緣影像。由於僅透過計算模組370，便可得知欲傳送 至目標像素Pix的目前影像成份F0—yc是否為一邊緣影 像’故在其它實施例中，可省略計算模組39〇。 在本實施例中，計算模組390包括平均器391以及邊 緣判斷器393。平均器391平均欲傳送至目標像素ρίχ的目# 前影像成份F0—yc及第二先前影像成份打―八。邊緣判斷 器393對平均态391的平均結果，進行邊緣判斷處理，用 以產生計算結果Sec在本實施例中，平均器391所產生的 平均結果係為欲傳送至目標像素pix的目前影像成份 F〇_yc及第二先前影像成份F2—yc的亮度成份的平均值 F〇2_y 。 在一可能實施例中，判斷模組35〇先根據計算結果Sd 及Sc2 ’產生一判斷結果sD],接著再根據計算結果Sc3及 12 201141243 sC4 ’定義出-第—預設值’然後將判斷結果s⑴與第一預 設值作比較’最後再根據比較結果，衫使處理單元⑴〇 或21〇)輸出，理信號或Sp2」D予目標像素❿。 在另貝靶例中，只要藉由計算模組370的計算結果 sC3’便可得知欲傳駐目標像素pix的目前f彡像成份抑％ 是否為-邊緣影像。因此，在此射，雌模組35〇只根 據計算模組370的計算結果，產生—第―預設值，再將第 一預設值與判斷結果SD1作比較。 在其它實施例中，可事先預設該第一預設值。在此例 =’判斷模組350可根據計算結果&3及&(或僅根據計 异結果Sc3)，調整事先預設的第一預設值，再將調整後的 第一預設值與判斷結果Sdi作比較，用以決定使處理單元 (110或210)輸出處理信號Spi_2D或Sp2—3D。 舉例而言，若計算模組370及39〇均計算出一邊緣時， 則判斷模組350減小第一預設值，使得判斷結果s⑴大於 第一預設值。因此，處理單元110對欲傳送至目標像素pix 的目前影像成份F0_yc進行2D梳型濾波處理。 相反地，若計算模組370及390並未計算出一邊緣時， 則判斷模組350增加第一預設值，使得判斷結果小於 第一預設值。因此，處理單元110對欲傳送至目標像素pix 的目前影像成份F0_yc進行3D梳型濾波處理。 第4圖為本發明之控制單元之另一可能實施例。第4 圖與第3圖相似’不同之處在於，第4圖多了統計模組46〇。 統計模組460對影像信號VlN1進行統計。判斷模組43〇可 根據統計模組460的統計結果，定義或調整一第一預設值。 13 201141243The 2D comb filter 211 performs a 2D comb type filtering process on the current image component F0_yc to be transmitted to the target pixel pix to generate a processing signal Sp 〇 2 〇. The 3D comb filter state 213 performs a 3D comb filter process on the current image component F0_yc to be transmitted to the target pixel pix to generate a processing signal Spud. The switch module 215 receives the processed signal %]1 and the heart 2 buckle, and determines the output processing signal or SP2_3D to the target pixel pix according to the output signal of the determining module 235. In the present embodiment, the determination module 235 generates a determination result ^ according to the calculation result Sg] generated by the calculation modules 23A and 233. When it is judged that =SD1 is larger than a first preset value, the mode is judged so that the switch mode j(7)' is controlled to output the processed signal Sp]_2D. When the judgment result s(1) is smaller than the first: = the module 235 controls the switch module - to make the output type 2 = in the example, the control unit 230 starts to operate, and the 1 and 3D comb chopper 213 starts to The current image component to be transferred to the 〇 χ χ 〇 〇 〇 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 她 = = = = = = = = = = = = = = = = = = = = = and. Since the calculation modules 310 and 330 of the control unit 30 and the determination module 350 are similar to the calculation modules 131 and 133 of the first figure and the determination module 135, they will not be described again. In Fig. 3, the calculation module 370 performs edge detection processing on the current image component F〇_yc to be transmitted to the target pixel Pix to generate a calculation result SC3. In the present embodiment, the calculation module 370 processes (for example, high-frequency filtering processing) the luminance component F0__y of the current image component F〇_yC to be transmitted to the target pixel pix to generate edge information. The calculation module 390 performs edge determination processing on the basis of the current image appeal component F0-yc and the second previous image component F2_yc' to be transmitted to the target pixel Pix to generate the calculation result SC4. By calculating the module 39〇, it is more reliable to transmit to the target pixel ρ [χ the current image component is - edge image. Since the current image component F0-yc to be transmitted to the target pixel Pix is known to be an edge image only through the calculation module 370, in other embodiments, the calculation module 39 can be omitted. In the present embodiment, the computing module 390 includes an averager 391 and an edge determiner 393. The averager 391 averages the destination image component F0-yc and the second previous image component to be transmitted to the target pixel ρίχ. The edge determiner 393 performs an edge determination process on the average result of the average state 391 for generating the calculation result Sec. In the present embodiment, the average result produced by the averager 391 is the current image component F to be transmitted to the target pixel pix. The average value F〇2_y of the luminance components of 〇_yc and the second previous image component F2-yc. In a possible embodiment, the determining module 35 first generates a determination result sD according to the calculation results Sd and Sc2', and then defines a -first-preset value according to the calculation results Sc3 and 12 201141243 sC4 'and then judges The result s(1) is compared with the first preset value. Finally, according to the comparison result, the shirt outputs the processing unit (1) or 21〇, and the signal or Sp2"D is given to the target pixel ❿. In the other example, the calculation result sC3' of the calculation module 370 can be used to know whether the current image component of the target pixel pix is %-edge image. Therefore, in this shot, the female module 35 〇 generates a -first preset value based on the calculation result of the calculation module 370, and compares the first preset value with the judgment result SD1. In other embodiments, the first preset value may be preset in advance. In this example, the judgment module 350 can adjust the preset preset value according to the calculation result & 3 and & (or only according to the difference result Sc3), and then adjust the first preset value. The comparison with the determination result Sdi is used to determine whether the processing unit (110 or 210) outputs the processing signal Spi_2D or Sp2 - 3D. For example, if both the computing modules 370 and 39〇 calculate an edge, the determining module 350 decreases the first preset value, so that the determination result s(1) is greater than the first preset value. Therefore, the processing unit 110 performs a 2D comb filtering process on the current image component F0_yc to be transmitted to the target pixel pix. Conversely, if the computing modules 370 and 390 do not calculate an edge, the determining module 350 increases the first preset value so that the determination result is less than the first preset value. Therefore, the processing unit 110 performs a 3D comb filtering process on the current image component F0_yc to be transmitted to the target pixel pix. Figure 4 is another possible embodiment of the control unit of the present invention. Figure 4 is similar to Figure 3'. The difference is that Figure 4 has more statistical modules 46〇. The statistics module 460 performs statistics on the image signal V1N1. The determining module 43 can define or adjust a first preset value according to the statistical result of the statistical module 460. 13 201141243
預設值。 在一可能實施例中， 的統計結果、計算結果s 在另一可能實施例中，判斷模組430具有事先設定的第一 預設值。在此例中，判斷模組430可根據統計46〇的 統計結果、計算結果So及Sc#，適當地調整第—預嗖值。 在本實施例中，統計模組460可對影像信號v_中的 圖框與圖框之間（或圖場與圖場之間）的絕對差值（Solute differences)’進行長方圖（histogram)統計，但此揭露並非用 以限制本發明。在其它實施例中，本領域之技術人員可利鲁 用其它的統計方式。另外，由於影像信號的絕對差值計算， 為本領域人士所深知，故不再贅述。 第5圖為影像k號V1N1的絕對差值histogram統計示意 圖（如二張相鄰圖框間像素點對點的絕對差值）。hist〇gram 統片值可用來輔助判斷移動現象。藉由hist〇gram統計值， 可知知目刖整體圖框是偏向靜止或是有在移動。即使有雜 訊(noise)影響，也可以正確判斷整張圖框是否有移動。 上述所提到的靜止及移動的判斷，係與上一圖框的比 ⑩ 較結果。以第6圖所示的圖框為例，若目前圖框(F+F〇)與 先前圖框(F1+F2)的差異過大時，表示目前圖框係對應到一 移動後的圖框。相反地，若目前圖框及先前圖框的差值不 大時，表示目前圖框對應到一靜止圖框，也就是目前圖框 所對應的畫面與先前圖框所對應的畫面相同。另一統計方 式亦可取二張相鄰的奇圖場或偶圖場（即同奇偶場， same-parity field)進行統計。例如取奇圖場F、F1或偶圖場 F〇、F2進行絕對差值histogram統計。 14 201141243 在第5圖中（靜止晝面的統計結果），根據統計最多的位 置，便可知道雜訊有多大。另外，若3D差值大於20的影 像成份的數量幾近零，則表示該圖框是靜止圖框。反之， 則表示圖框有在移動。再者，觀察3D差值較大的影像成 份的數量，便可得知整張圖框是否有部分物件被移動。 倘若藉由histogram統計出，整張圖框是靜止的現象， 則與判斷結果SD1比較的該第一預設值就可以設定成靜止 適用的數據，因此，histogram統計值可用以控制該第一預 Φ 設值的大小。 在一可能實施例中，根據一預設圖框，便可定義出一 第一預設值。在其它實施例中，可利用其它的方式，定義 出第一預設值。 請參考第1圖，判斷模組135將該第一預設值與判斷 結果SD1作比較後，便可判斷出欲傳送至目標像素Pix的目 前影像信號F0_yc係為靜止或移動像素（與該預設值相比 較）。以上所訴的實施例係依照目前圖框/圖場中，欲傳送至 • 每一點像素的影像成份的情況（靜止或移動影像成份），而 採用2D或3D梳型濾波方式來處理該影像成份，也就是 說，當該影像成份被判斷為靜止影像成份時，則採用3D 梳型濾波方式來處理此影像成份，否則以2D梳型濾波方 式來處理此影像成份。 雖然本發明已以較佳實施例揭露如上，然其並非用以 限定本發明，任何所屬技術領域中具有通常知識者，在不 脫離本發明之精神和範圍内，當可作些許之更動與潤飾， 因此本發明之保護範圍當視後附之申請專利範圍所界定者 15 201141243 為準。default value. In a possible embodiment, the statistical result, the calculation result s. In another possible embodiment, the determination module 430 has a first preset value set in advance. In this example, the determination module 430 can appropriately adjust the first pre-depreciation value based on the statistical result of the statistics 46〇, the calculation results So and Sc#. In this embodiment, the statistics module 460 can perform a histogram on the absolute difference (Solute differences) between the frame and the frame in the image signal v_ (or between the field and the field). Statistics, but this disclosure is not intended to limit the invention. In other embodiments, those skilled in the art can use other statistical methods. In addition, since the absolute difference calculation of the image signal is well known to those skilled in the art, it will not be described again. Figure 5 is a statistical diagram of the absolute difference histogram of the image k number V1N1 (such as the absolute difference of pixel points between two adjacent frames). The hist〇gram system value can be used to assist in determining the movement phenomenon. By using the hist〇gram statistic, it can be known that the overall frame is biased to be stationary or moving. Even if there is noise, it is possible to correctly judge whether the entire frame has moved. The above-mentioned judgments of stillness and movement are compared with the results of the previous frame. Taking the frame shown in Fig. 6 as an example, if the difference between the current frame (F+F〇) and the previous frame (F1+F2) is too large, it indicates that the current frame corresponds to a moved frame. Conversely, if the difference between the current frame and the previous frame is not large, it indicates that the current frame corresponds to a still frame, that is, the picture corresponding to the current frame is the same as the picture corresponding to the previous frame. Another statistical method can also take two adjacent odd or even fields (that is, the same-parity field) for statistics. For example, take the odd field F, F1 or the even field F〇, F2 to perform absolute difference histogram statistics. 14 201141243 In Figure 5 (statistical results of the stationary surface), you can know how big the noise is based on the most statistically located position. In addition, if the number of image components having a 3D difference greater than 20 is nearly zero, it indicates that the frame is a still frame. Otherwise, it means that the frame is moving. Furthermore, by observing the number of image components having a large difference in 3D, it is possible to know whether or not some of the objects are moved in the entire frame. If the whole frame is static by the histogram, the first preset value compared with the judgment result SD1 can be set to the data that is statically applicable. Therefore, the histogram statistical value can be used to control the first pre-predetermined data. Φ Set the size of the value. In a possible embodiment, a first preset value can be defined according to a preset frame. In other embodiments, the first predetermined value may be defined in other manners. Referring to FIG. 1, the determining module 135 compares the first preset value with the determination result SD1, and then determines that the current image signal F0_yc to be transmitted to the target pixel Pix is a stationary or moving pixel (with the pre- Set the value to compare). The above-mentioned embodiments are based on the current frame/field, where the image component is to be transmitted to each pixel (still or moving image component), and the image component is processed by 2D or 3D comb filtering. That is to say, when the image component is judged to be a still image component, the image component is processed by a 3D comb filter, otherwise the image component is processed by a 2D comb filter. Although the present invention has been disclosed in the above preferred embodiments, it is not intended to limit the invention, and any one of ordinary skill in the art can make some modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention is defined by the scope of the patent application, which is incorporated herein by reference.
16 201141243 【圖式簡單說明】 第1圖為本發明之梳型濾波裝置之一可能實施例。 第2圖為本發明之梳型濾波裝置之另一可能實施例。 第3圖為本發明之控制單元之另一可能實施例。 第4圖為本發明之控制單元之另一可能實施例。 第5圖為影像信號VIN]的絕對差值histogram統計示意 圖。 第6圖為影像信號VW1示意圖。 • 【主要元件符號說明】 100、200 :梳型濾波裝置； 110、 210 :處理單元； 130、230、300、400 :控制單元； 111、 211 : 2D梳型濾波器； 113、213 : 3D梳型濾波器； 13卜 133、23卜 233、310、330、370、390、410、420、 440、450 :計算模組； ^ 135、235、350、430 :判斷模組； 132、134 :内部2D梳型濾波器； 136 :計算器； 137 :混合器； 215 :開關模組； 391 :平均器； 393 :邊緣判斷器； 460 :統計模組。 1716 201141243 [Simple description of the drawings] Fig. 1 is a possible embodiment of the comb filter device of the present invention. Figure 2 is another possible embodiment of the comb filter device of the present invention. Figure 3 is another possible embodiment of the control unit of the present invention. Figure 4 is another possible embodiment of the control unit of the present invention. Figure 5 is a statistical diagram of the absolute difference histogram of the image signal VIN]. Figure 6 is a schematic diagram of the image signal VW1. • [Main component symbol description] 100, 200: comb filter device; 110, 210: processing unit; 130, 230, 300, 400: control unit; 111, 211: 2D comb filter; 113, 213: 3D comb Type filter; 13 133, 23 233, 310, 330, 370, 390, 410, 420, 440, 450: computing module; ^ 135, 235, 350, 430: judgment module; 132, 134: internal 2D comb filter; 136: calculator; 137: mixer; 215: switch module; 391: averager; 393: edge determiner; 460: statistical module. 17