591958 五、發明說明（l) ----—----- 【技術領域】 以it明為一種可調式臨界值之圖框種類偵測方法。係 以一可調的酢反#仏 序式圖框之判斷，：為$影像中的圖框是為交錯式圖框或循 錯誤判斷，以達到i避免因場差異過大所造成的圖框種類 【先前技術】“判斷圖框精確度之目的。 (.^ / A晝面的影像其係由每秒傳送3 0個圖框 你m 4所組成，進而產生連續的動態影像，每一圖框 (aNat.onal Telev1S1〇n eji—Committee ;NTSC為例，係傳送““条掃描線）， 尤疋母一圖框係利用1/30秒來傳送525條掃描線。 式圖述視訊的圖框目前可分為二種，分別可分為交錯 frame frame)以及循序式（Progressive me )圖框，其中交錯式圖框其掃描方式係 (top^f.eld) (bott^flerd 線，頂圖場為偶數的㈣線’底圖場則為奇數的奸 _ =後分成兩次分別掃描，如先掃描 = 圖場後產生一完整的圖框；而循序 口琢後再“底 ^ ^ ^ ^ ^ ^ ^ ^ # ^ ^ ^ ^ ( δρΓ^δ^ ® 勺掃描線，然後一線一線循序掃描。 斤有 如第一 （a )圖及第一 （b )圖所示八如炎a w 、 1 式序”框U之影像示意圖，由前：可知，由日if框 數婦描線）及底圖場⑽掃描線 五 發明說明（2) 圖方形影像100及三角形影像101其輪廓邊緣產 ，^、1像（一般稱為梳狀因數（comb factor ) 框11 G tL:對的畫面品質亦較差；而由於循序式 1由於為同時傳送整個圖框，因此其不會產生如第— a)圖中之梳狀因數。 與由前述可知由於梳狀因數的產生對於解析度有極大 Γ二图因此在輸出影像之前會進行一偵測流程，以偵测 :卜^為交錯圖框或是循序式圖才匡，若偵測為交錯圖 面於輸出之前會有一解交錯（deinterlacing 处4 \以將梳狀因數去除，以獲得更佳的影像輸出品 你右為循序式圖框，則不需經過解交錯步驟而直接輪出 像即可。 2、’先的偵/則流程係如第二圖所示，首先在步驟2 〇 1 ，設計數器為零，輸入一可計算出晝素梳狀因數值 faCt〇r )之圖框函式F (n )，接著進行步驟202， 金 晝素代入此圖框函式F (η )中，步驟203，對所有 ^素進行掃描以計算晝素之梳狀因數值（c〇mb fact〇r “ a 5梳狀因數值計算之演算式容後再.敘；步驟2 0 4中， 判斷每一畫素之梳狀因數值是否大於臨界值，若大於臨 值，則進入步驟205中，將計數值加i ;若梳狀因數值小°於 界值，則進入步驟206，判斷是否為最後一圖框晝素之 c狀因數，若疋，則進入步驟2 〇 7 ;若否，則重複步驟2 〇 3 至步驟20 6 ;步驟207，判斷計數器中數值是否小於臨界 值，若計數器中數值小於臨界值時，則於步驟2〇8判斷其 591958 五、發明說明（3) _ 圖框種類為循序式圖框，若計數器數值大於臨界值時 進入^驟209判斷其圖框種類為交錯式圖框、 偵測流程。 果 第圖所不係為在第二圖中步驟^ 梳狀因數值的圖框書素示咅R ^ ^ 旦畜具 圖中所示之χ(座標：，二，其中欲計算畫素30發生於 (PW value)為」，而位置，且其晝素值 為與晝素30在不同圖場之鄰近素晝素32 之畫素值為” b”、第二畫素32之旦值、]:晝素31 素30梳狀因數值的演算式為··息素值為e ，而其計算畫 梳狀因數（CombFactor ( γ 、、 -(b-e ) 2 5 ^ ) ) = (b-χ ) (e-x ) 雖然利用演算式算出每一 -設定的固^臨界值作比 ^的梳狀因數數值然後與 狀因數，最後再根據梳狀因數:^一，斷此晝素是否為梳 )可判斷出圖框種類，但3勺〜和數目（即計數器數目 因是由於梳狀因數的數目=$，方法亦會造成誤判。其原 (bottem field)的差里=$ (top field)與底場 例如第四圖所示，其大致略^ =關係，其差異的比例關係 說如果一交錯式圖框包含 正比的線性關係；也就是 麼場差異值較小，相對而古=動作或較低頻率的物件，那 影像内容較複雜時，例如^ ^ =因數數目亦會較少。但若 等較複雜的物件時，即使 包含有例如樹葉、碎石頭 即為循序式圖框），其 f場及底場間沒有時間差（亦 _ 、异出的梳狀因數的數目仍有可能 第6頁 591958 五、發明說明（4) 此，如果只使用一固定的臨界值作為判斷… Γ;=ί 2有可能會產生將一慢動作或較低頻率= 序式圖框誤判為一交伊犬圖框二决動作或兩頻率的循 】框未經過解交錯處理步驟即輪出，使影像輸：以式 *為解决上述因使用固定臨界值而造成圖框種類。 臨種類判斷的正確率提升，本發明提出；ΐ 框及循序式圖框判斷方法。 、曰式圖 【發明内容】 本叙明為一種可調式臨界值之圖框方 以影像中場差異的訊息作一初步的判斷作筆 差異值動態的調整計數罘砟χ^ . 1 …、後根據％ 種類是屬於交錯式圖框(win二i ’曰d中其圖框 的鋩β)之判斷，以屏除場差異所造成 ’曰^、斷，並達到提高判斷圖框種類精確度之目的。591958 V. Description of the Invention (l) ------------ [Technical Field] A method of detecting frame types with it as an adjustable threshold. The judgment is based on an adjustable 酢 ## sequential frame: the frame in the image is an interlaced frame or a wrong judgment, so as to avoid the type of frame caused by excessive field differences [Previous technology] "The purpose of judging the accuracy of the frame. (. ^ / A daytime image is composed of 30 frames per second and you m 4 to produce continuous dynamic images, each frame (aNat.onal Telev1S10n eji—Committee; NTSC, for example, is used to transmit "" scan lines), and the frame of the maiden is used to transmit 525 scan lines in 1/30 second. The picture frame of the video At present, it can be divided into two types, which are divided into interlaced frame frame and progressive me frame. The interlaced frame has the scanning method (top ^ f.eld) (bott ^ flerd line, top field). The base field for the even number of base lines is an odd number _ = and then divided into two separate scans, such as scanning first = a complete frame is generated after the field; and then “bottom ^ ^ ^ ^ ^ ^ ^ ^ ^ # ^ ^ ^ ^ (δρΓ ^ δ ^ ® spoon scan line, and then scan line by line. Sequentially as the first (a) picture and the first (B) The schematic diagram of the image of the pattern "Baoyanwaw" in the figure "Frame U", from the front: we can see that the line is drawn from the Japanese if box) and the scan field of the base map. (5) Explanation of the square image 100 and triangle image 101 have contour edges, ^, 1 image (commonly called comb factor). Frame 11 G tL: The picture quality of the pair is also poor; and because the sequential formula 1 is to transmit the entire frame at the same time, Therefore, it does not generate the comb factor as shown in the figure-a). As can be seen from the foregoing, the comb factor has a great resolution for the second image. Therefore, a detection process is performed before the image is output to detect : Bu ^ is an interlaced frame or a sequential image. If it is detected as an interlaced image, there will be a deinterlacing before the output (deinterlacing at 4 \ to remove the comb factor to get a better image output.) The right is a sequential frame, and you can directly rotate the image without going through the de-interlacing step. 2. The 'first detection / then flow' is shown in the second figure. First, at step 2 0, set the counter to zero. , Input one to calculate the value of daytime comb factor faCt〇r) Frame function F (n), and then proceed to step 202. Jin day prime is substituted into this frame function F (η). Step 203 scans all primes to calculate the comb factor value of day prime (c. mb fact〇r "a 5 The calculation formula of the numerical value of the comb factor is described later; in step 204, it is judged whether the comb factor value of each pixel is greater than the critical value, and if it is greater than the threshold value, the process proceeds to step In 205, add the count value to i; if the value of the comb-shaped factor is smaller than the threshold, go to step 206 to determine whether it is the c-shaped factor of the last day frame prime, if it is not, go to step 2 07; if If no, repeat step 2 03 to step 20 6; step 207, determine whether the value in the counter is less than the critical value, and if the value in the counter is less than the critical value, then judge 591958 in step 2 08 V. Description of the invention (3) _ The frame type is a sequential frame. If the counter value is greater than the critical value, enter step 209 to determine that the frame type is an interlaced frame and the detection process. If it is not shown in the second figure, the frame of the comb factor value in the second figure is shown in the picture book element 咅 R ^ ^ (coordinates :, two, where the pixel 30 to be calculated occurs in the vise figure) Where (PW value) is ”, and the position is, and its daytime prime value is adjacent to the daytime prime 30 in a different field. The pixel value of the primetime prime 32 is“ b ”, the value of the second pixel prime 32,] : The calculation formula of the combing factor value of day prime 31 and prime 30 is · · the value of the pectin is e, and the calculation of the combing factor (CombFactor (γ,--(be) 2 5 ^)) = (b-χ) (ex) Although the combing factor value of each -set fixed ^ critical value is calculated using the calculation formula and then compared with the shape factor, finally according to the combing factor: ^ 1, whether this day element is a comb) can be judged Type of picture frame, but 3 scoops ~ and the number (that is, the number of counters is due to the number of comb factors = $, the method will also cause misjudgment. The difference between the original (bottem field) = $ (top field) and the bottom field For example, as shown in the fourth figure, the relationship is roughly ^ =. The proportional relationship of the difference is that if an interlaced frame contains a linear relationship that is proportional; Ancient = action or lower frequency objects, when the image content is more complicated, for example, ^ ^ = the number of factors will be less. But if more complex objects, even if it contains leaves, broken stones, for example, it is sequential (Picture frame), there is no time difference between the f field and the bottom field (also _, the number of out-of-order comb factors is still possible. Page 6 591958 V. Description of the invention (4) Therefore, if only a fixed critical value is used as Judgment ... Γ; = ί 2 may produce a slow motion or a lower frequency = the sequence frame is misjudged as a one-shot Idog frame two-shot action or two-frequency cycle] The frame does not go through the de-interlacing process step Output, make the image lose: use formula * to solve the above-mentioned frame types caused by the use of a fixed critical value. The accuracy of the type judgment is improved, the present invention proposes; ΐ frame and sequential frame frame judgment method. Summary of the Invention This description is a kind of adjustable threshold value frame to make a preliminary judgment based on the information of the field difference in the image to make a dynamic adjustment of the difference value 罘 砟 χ ^. 1…, and then according to the% type belongs to Staggered frame The judgment of 图 β) of the picture frame in d is to remove the '^ and break caused by the field difference, and achieve the purpose of improving the accuracy of the picture frame type.
測方ί ί ί 士述目!!、，本發明可調式臨界值之圖框種類搞 f ' ’、"又计數器為零，並時間場η時輸入圖框一函式F (fraffle) 白^ # ^異上限值（Fi-th)比較；當場差異值小於設定 動二66二Ϊ限值i F 1 ~ U )時，即可根據場差異值的訊息 @ ^伯设定計數器臨界值；然後算出圖框中大於梳狀因數 " 之梳狀因數數目並與該設定之計數器臨界值比較； 第7頁 591958 五、發明說明（5) '—'一" 並判斷其圖框種類係屬於交錯式圖框或是循序式圖框；以 決定是否進行解交錯晝素的步驟。 【實施方式】 第五圖係為本發明一較佳實施例可調式臨界值之圖框 種類偵測方法之一流程示意圖，係包括下列步驟： ° 首先，進行步驟501，設定計數器為零，輸入一圖框函式f (η )，由於此圖框函式在習用技術中已述及，在此不^ 贅敘。接著進行步驟5 02，計算出圖框（frame)在時間 n時的場差異值，其場差異值演算式可為： 場差異值（FieldDiff) 其中，Μ及N分別代表圖框（frame )的高度及寬度， F(x，2y, η)及F(x，2yf 1，η)則分別代表其頂圖場及底圖場 晝素值。 當算出場差異值後，即進入步驟5〇3，將算出之場差 異值與一設定之場差異上限值（F丨—th )比較；若場差異 值大於或等於設定的場差異上限值時，即進入步驟5丨2， 判定其為交錯式圖框；若場差異值小於設定的場差異上限 值（Fi-th )時，即進入步驟5〇4，動態的調整計數器臨界 值，其凋整方式谷後再敘；接著，在步驟5 〇 5中將圖框所 有晝素代入圖框函式F ( η )中，並計算出所有晝素的梳狀 因數值506 ;於步驟507中判斷梳狀因數值是否大於或等於 梳狀因數臨界值（combTH)，若梳狀因數臨界值大於或' 等於梳狀因數臨界值時，即將計數器加j (步驟5 〇 8 )，若测 方 ί ί 述 Description, the frame type of the adjustable threshold value of the present invention is f '', " and the counter is zero, and a function F (fraffle ) White ^ # ^ Different upper limit value (Fi-th) comparison; when the field difference value is less than the set limit of two two 66 two threshold i F 1 ~ U), you can set the counter based on the message of the field difference value @ ^ 伯Critical value; then calculate the number of comb factors greater than the comb factor " in the frame and compare it with the set threshold of the counter; page 7 591958 5. Description of the invention (5) '—' 一 " and judge the graph The frame type belongs to an interlaced frame or a sequential frame; to determine whether to perform the step of deinterlacing the day element. [Embodiment] The fifth diagram is a schematic flow chart of a method for detecting the frame type of the adjustable threshold according to a preferred embodiment of the present invention. The method includes the following steps: First, go to step 501, set the counter to zero, and enter A frame function f (η). Since this frame function has been described in the conventional technology, it will not be repeated here. Then step 5 02 is performed to calculate the field difference value of the frame at time n. The field difference value calculation formula can be: FieldDiff where M and N represent the frame Height and width, F (x, 2y, η) and F (x, 2yf 1, η) represent the day field values of the top field and the bottom field, respectively. When the field difference value is calculated, it proceeds to step 503, and the calculated field difference value is compared with a set upper limit value (F 丨 —th) of the field difference; if the field difference value is greater than or equal to the set upper limit of the field difference If the value of the field difference is less than the set field difference upper limit value (Fi-th), then enter the step 5104 to dynamically adjust the threshold value of the counter. , Its decay mode will be described later; then, in step 505, all frame elements are substituted into frame function F (η), and the comb factor values of all elements are calculated 506; in step In 507, it is determined whether the comb factor value is greater than or equal to the comb factor critical value (combTH). If the comb factor threshold is greater than or equal to the comb factor critical value, the counter is incremented by j (step 5 08).
第8頁 591958 五、發明說明（6) 梳狀因數臨界值小於梳狀因數臨界值時，則進入步驟 509，步驟5〇9，判斷是否為最後一畫素，若是，即進入步 驟510 ;若否，則重複步驟50 6，繼續計算下一梳狀因數 值二^驟510，判斷計數器數值是否小於或等於調整之計 數杰g品界值，若是，則判定其為循序式圖框，若否，則判 定其為交錯式圖框512，最後，結束偵測流程。 在步驟5 0 4中’其計數器臨界值的動態調整，其一實 施例例如第六圖之圖式，請一併參閱第四圖，由於由第四 圖中可知場差異值的大小與梳狀因數數目的大小略呈一線 性的比例關係，因此，可於將場差異值再細分為數個場差 異數值以區隔出數個範圍，在本實施例中係分為四個，分 別為第一場差異值di、第二場差異值d2、第三場差異值d3 及第四場差異值d4 ;其中第一場差異值即為前述的場差異 上限值（Fi-th)，當計算出的場差異值介於第一場差異 值dl及第二場差異值d2之間的範圍時，則其計數器的臨界 值動態調整至第一預設計數器臨界值（c〇unter —)， 當計算出的場差異值介於第二場差異值d2及第三場差異值 d 3之間的範圍時，則其計數器的臨界數值動態調整至第二 預設計數器臨界值（Counter-th2);當計算出的場差異 值介於第三場差異值d3及第四場差異值d4之間的範圍時， 則其計數器的臨界值動態調整至第三預設計數器臨界值 (Counter-th3 );若計算出的場差異值介於第四場差異 值d4及0之間的範圍時，則其計數器的臨界值動態調整至 第四預設計數器臨界值（Counter-th4 )。Page 8 591958 V. Description of the invention (6) When the critical value of the comb factor is less than the critical value of the comb factor, the process proceeds to step 509 and step 509 to determine whether it is the last pixel. If so, the process proceeds to step 510; if If not, repeat step 50 6 and continue to calculate the next combing factor value 2 ^ 510 to determine whether the counter value is less than or equal to the adjusted count value. If it is, then it is determined as a sequential frame. If not, , It is determined to be an interlaced frame 512, and finally, the detection process is ended. In step 504, the threshold value of the counter is dynamically adjusted. An example of the embodiment is the diagram of the sixth diagram. Please refer to the fourth diagram together, because the magnitude and comb shape of the field difference value can be known from the fourth diagram. The number of factors has a slightly linear proportional relationship. Therefore, the field difference value can be further subdivided into several field difference values to separate out several ranges. In this embodiment, it is divided into four, which are the first The field difference value di, the second field difference value d2, the third field difference value d3, and the fourth field difference value d4; where the first field difference value is the aforementioned field difference upper limit value (Fi-th), and when calculated, When the value of the field difference is between the first field difference value dl and the second field difference value d2, the threshold value of the counter is dynamically adjusted to the first preset counter threshold value (c〇unter —). When the field difference value is in a range between the second field difference value d2 and the third field difference value d 3, the threshold value of the counter is dynamically adjusted to the second preset counter threshold value (Counter-th2); The calculated field difference is between the third field difference d3 and the fourth field difference d When the range is between 4, the threshold value of its counter is dynamically adjusted to the third preset counter threshold value (Counter-th3); if the calculated field difference value is in the range between the fourth field difference value d4 and 0 , The threshold value of the counter is dynamically adjusted to a fourth preset counter threshold value (Counter-th4).
591958 五、發明說明（7) _ 值得一提的是，將該場差異值蛊一 可提供一快速的圖框種類辨別方式，、—θ共上限值比較 差異上限值，須經由第五圖步驟5H:差異:未超過場 框為交錯式圖框或為循序式圖框；〇來判斷此圖 上限值，則可直接辨別為交錯式；：穷差異超過場差異 驟504至步驟510之步驟。 °王不需經由第五圖步 以上為本發明可調式臨界值 的場差異值與-場差異上限值作 e )在時間η時 再動態的調整計數器的臨界值，藉 土：：驟’然後 或循序式圖框的判斷，以提古^ 。對為交錯式圖框 综上所述，充份顯正確率。 之利用價值，且為目…施之進步性，極具產業 合發明專利之要件，；依法斤J見之新發明’完全符 能以所巧本發明之較隹實施例而已，當不 分成四個臨界值，更：：：U ’例如其計數器不僅只能 即大凡依本發明申心b々而要为成更多或更少個臨界值。 應仍屬於本發明^專利範圍所作之均等變化與修飾，皆 鑑，並祈惠准ί:至函範圍内’謹請貴審查委員明 591958 圖式簡單說明 第一 （a )圖係為交錯式圖框之影像輸出示意圖； 第一（b )圖係為循序式圖框之影像輸出示意圖； 第二圖係為習用技術偵測圖框種類方法之流程示意圖； 第三圖所示係為計算一晝素其梳狀因數值的圖框晝素示意 圖， 第四圖係為梳狀因數的數目與場差異值之示意圖； 第五圖係為本發明可調式臨界值之圖框種類偵測方法其較 佳實施例之一流程示意圖；及 第六圖係為本發明實施例將場差異值再細分為四個臨界值 之示意圖。 【符號說明】 1 0交錯式圖框； 1 0 0方形影像； 1 0 1三角形影像； 11循序式圖框； 30晝素； 3 1第一鄰近晝素； 32第二鄰近晝素； dl第一場差異值； d2第二場差異值； d3第三場差異值； d4第四場差異值； counter-thl第一預設計數器臨界值； 。〇1111七61'-1:112第二預設計數器臨界值，·591958 V. Description of the invention (7) _ It is worth mentioning that the field difference value can provide a fast way to identify the type of frame. The difference between the upper limit of -θ and the upper limit of the difference must pass the fifth Step 5H in the figure: Difference: The frame that does not exceed the field frame is an interlaced frame or a sequential frame; 〇 Judging the upper limit of the figure, it can be directly identified as an interlaced frame: Poor difference exceeds the field difference. Step 504 to step 510 The steps. ° The king does not need to go through the fifth figure and above. The field difference value and the -field difference upper limit value of the adjustable threshold of the present invention are used as e.) The threshold value of the counter is dynamically adjusted at time η. Then, or the judgment of the sequential frame, to mention Tigu ^. For the staggered frame, the accuracy rate is fully shown. The use value is for the purpose of ... the progressiveness of the application, which is very important for industrial and invention patents; the new invention according to the law is completely consistent with the comparative example of the present invention, and it is not divided into four. For example, the threshold value can be more than :: U '. For example, the counter can not only be a threshold value that can be set to be more or less according to the present invention. It should still belong to the present invention ^ Patent scope for equal changes and modifications, all learned, and pray for approval: Within the scope of the letter 'I would like to ask the reviewing committee member 591958 Schematic description of the first (a) diagram is staggered Schematic image output of the frame; The first (b) image is a schematic image output of the sequential frame; the second image is a flow chart of the method of detecting frame types by conventional techniques; the third image is a calculation Schematic diagram of the daytime prime and its comb factor values. The fourth diagram is a schematic diagram of the number of field factors and field difference values. The fifth diagram is the frame type detection method of the adjustable threshold of the present invention. A schematic flowchart of one of the preferred embodiments; and the sixth diagram is a schematic diagram of subdividing the field difference value into four critical values according to the embodiment of the present invention. [Symbol description] 10 interlaced frames; 100 square images; 101 triangular images; 11 sequential frames; 30 day elements; 3 1 first adjacent day elements; 32 second adjacent day elements; One field difference value; d2 second field difference value; d3 third field difference value; d4 fourth field difference value; counter-thl first preset counter threshold value; 〇1111 Seven 61'-1: 112 Second preset counter threshold, ·
第11頁 591958Page 591958