經濟部中央橾準局員工消费合作社印製 318317 A7 B7五、發明説明（2) 〔產業上之利用領域〕 本發明係關於將被編碼之動晝予以解碼之動畫解碼裝 s，特別是在解碼中有錯誤被檢測出時，能夠不使輸出之 圖像品質降低，而能將錯誤之影響排除者。 〔先前之技術〕 一般而言，被數位化之圚像資料之保存，需要膨大之 記憶容量。而動畫時需要更大的容量。於是，爲了將動畫 編碼壓縮稹蓄，有各式各樣的手法。 MPEG (Moving Picture Expert Group)視頻之 規格也是其中之一，可使彩色動晝之資料記憶容素顯著地 減低。其記億容量減低的手法之特徵爲： (1 )使用離散餘弦變換之二次元的壓縮。 (2 )依據時間方向的預測之動補償。 (3 )使用可變長編碼之平均信息量。 有關於MPEG視頻規格之說明於「I SO/I EC 1 1 1 7 2 — \ / 2 / 3 ! 'Information Technology-coding of Moving Pictures and Associated Audio for Digital Storage Media at up to about 1.5 Mbit/ s’，Internatiional Standard, 1993」有記載。 但是在此規格，正常地將圚像編碼之順序作了記述， 但是對於發生錯誤時之情況沒有任何之說明。又，詳細之 編碼之順序，參照上述規格。 又，關於錯誤發生時之對處方法，已知的有日本特開 紙&尺度適用中國國家橾準（CNS )八4規格（210X297公釐）-4 - ~~ (請先閲讀背面之注意事項再填寫本頁) 318317 A7 經濟部中央橾準局員工消费合作社印裝 B7五、發明説明（2) 平6 — 9 8 3 1 3號，在比發明中，在能夠成爲參照圖像 之圖像之編碼時，有不能迴避的錯誤發生時，由於係將發 生錯誤之參照圖像，以過去的參照圖像來代用予以對處， 所以有在其後所表示之畫質劣化之問題點。 此處，在解碼時發生錯誤的原因有， (1 )於保存CD等之資料之媒體，有物理上的破損 ，讀出之資料不能修復，即損壞時； (2 )傅送中的信號有千擾所造成之雜音進入，而使 資料變化時； (3 )於編碼時，由於使用規格外之方法，而使本來 不會出現之資料被輸入時；等之情況，像這種情況時，解 碼的繼縯成爲不可能，或使被解碼之圖像與原圖像大爲不 同。 〔發明所欲解決之課題〕 MP E G視頻規格之特徵之一的時間方向的預測之資 料壓縮，並不是以圖像資料本身作爲資料傳送或保存於記 憶體，係以過去或未來的圚像與現在之圖像之差成分作爲 資料傳送或保存於媒體，解碼時以過去或未來之圖像作爲 « 參照圊像，使用差成分，使現在的圖像再現之技術。 所以，不只限於MPEG視頻之規格，使用時間方向 之預測，使達成被編碼之資置的壓縮之技術，於將應參照 圖像解碼時，資料破壤等發生，圖像破損時，誤差會向參 照其後之圖像的圖像，或甚至參照該圖像之圖像俥播，會 (請先閔讀背面之注意事項再填寫本頁) .裝· 訂 線 本紙張尺度適用中國國家標準（CNS)A4说格（210X297公釐）-5 - 經濟部中央樣準局員工消费合作社印裝 318317 A7 ___B7五、發明説明（3) 造成動耋畫質之向下。 例如MP E G視頻規格中，被編碼之圖像的型有以下 數種。 (1 )不使用時間方向之預測的型之圖像。 (2 )使用順方向之時間預測之p型之圖像。 ---------- (3 )使用雙方向之時間預測之互型之圖像。 此處所謂的順方向之時間預測，係意味從過去之圖像 預想動作者，所謂雙方向係意味著從過去與未來的雙方之 圖像預想動作者。 I型之圖像，由於不使用時間方向之預測，於解碼時 ，不需要參照圖像。P型之圖像，由於係使用順方向之時 間預測，以過去的圚像作爲參照圖像，使用與參照圖像之 差成分，進行解碼。此處成爲參照圖像者，係剛被解碼之 I型或P型之圖像。B型之圖像，係使用以過法與未來的 雙方之圖像作爲參照圖像。此處被參照者，係先被輸出之 I型或P型之圖像，及先被解碼但輸出較後之I型或P型 之圖像。 圚1 0係表示依據MPEG視頻之解碼裝置中，解碼 之順序即被輸入之圖像的順序與輸出之圖像之順序之例之 圖。圚1 0中，6 1係被解碼之圚像列，6 1 a〜6 In 係個個被解碼之圖像，6 2係被输出之圖像列，6 2 a〜 6 2 m係表示各被輸出之圇像。又，付記於各圖像之符‘號 I、P、B，係表示其圖像之型，接著圖像之型而被付記 之號碼，係表示被输出之順序。又，圇1 0所示之圖像列 (請先閱讀背面之注意事項再填寫本頁) .裝· 訂 線 本紙張尺度逍用中國國家梯準（CNS)A4規格（ 210X297公釐）-6 - 經濟部中央棣準局員工消费合作社印裝 A7 __B7_五、發明説明（4 ) 爲一例，並不一定要如圖1 〇所示之順序使各型之圖像資 料被輸入，例如動作比較少的動畫像時，B型之圖像資料 連績者也有。相反的，動作比較激列之動畫像時，I型之 圖像資料被連績输入者也有。 在此列中，首先不需要參照圖像之參照6 1 a ( I 1 )被解碼。接著以圖像6 1 a作爲參照圖像，使圖像 6 1 b ( P 4 )被解碼。接著，以圖像6 1 a，6 1 b作 爲參照圖像，圖像6 1 c，同樣的以圖像6 1 a ，6 1 b 作爲參照圖像，參照6 1 d (B2 ，B3 )被依序解碼。 其次，以圖像6 1 b (P4 )作爲參照圖像，使圖像 6 1 e ( P 7 )被解碼。 另一方面，在本例中，於圚像6 1 c (B 2 )的解碼 被開始之後，输出開始。圚像6 2 a，係初期狀態，不是 有意的圖像。此處，圖像61b (P4)係比圖像61c ，6 Id (B2、B3)先被解碼，但是輸出是成爲在圖 像 61c，61d(B2、B3)之後。 以這樣的順序進行解碼時，例如圖像6 1 a (II) 之解碼時，如有錯誤發生而被錯誤地解碼時，以圖像 6 1 a ( I 1 )作爲參照圖像使用之圖像6 1 b ，6 1 c ，6 Id (P4，B2，B3)也具有錯誤，於是錯誤會 傳達至以圖像6 1 b ( P 4 )作爲參照圚像而使用之圖像 61e (P7)。如此，不需要參照圖像之I型之圖像 61k (Im)出現，更者直到僅只以圖像61k (Im )作爲參照圖像之參照61n (Pm+3)被解碼爲止， 請 閱 讀 背 1¾ 之 注 意 事 項 再 填 本 頁 裝 訂 線 本紙張尺度逍用中國國家梯準（CNS)A4規格（ 210X297公釐）-7 - 318317 A7 B7 五、發明説明（5 ) 會受到錯誤的影響，結果會導致動畫之畫質的低下。 本發明之目的，係提供於使用時間方向之預測之動畫 之編碼方式中，編碼化時有任何之錯誤被檢測出時，排除 其錯誤之影響，可避免畫質之品質降低之動畫解碼技術。 本發明之上述目的及其它之目的及新穎之特徵，可從 本說明書之記載及圖式了解。 〔解決問題之手段〕 本發明之上述目的，係於編碼時，於錯誤被檢出時刻 ，調査解碼中的畫像之型，而其爲作爲參照圈像所使用之 I型或P型時，直到不需要參照圖像之I型之圖像再出現 爲止，跳過讀取輸入資料，同時將出現的I型之圖像解碼 之後，再直到I型或P型之圖像出現爲止，再跳過讀取输 入資料之演算法，本發明即藉由採用此演算法來達成其目 的者。 〔作用〕 經濟部中夹樣準局員工消費合作社印裝 (請先閱讀背面之注意事項再填寫本頁)Printed 318317 A7 B7 by the Employees Consumer Cooperative of the Central Bureau of Economic Affairs of the Ministry of Economy V. Description of the invention (2) [Utilization field in the industry] The present invention relates to an animation decoding device that decodes the encoded moving day, especially in decoding When an error is detected, it is possible to eliminate the influence of the error without reducing the quality of the output image. [Previous Technology] Generally speaking, the storage of digitized image data requires an expanded memory capacity. The animation needs a larger capacity. Therefore, in order to encode and compress the animation, there are various methods. MPEG (Moving Picture Expert Group) video specifications are also one of them, which can significantly reduce the data memory capacity of color moving days. The characteristics of its method of reducing the capacity in billions are: (1) Compression of the second element using discrete cosine transform. (2) Predicted motion compensation based on the time direction. (3) The average amount of information using variable length coding. The description of the MPEG video specifications is in "I SO / I EC 1 1 1 7 2 — \ / 2/3! 'Information Technology-coding of Moving Pictures and Associated Audio for Digital Storage Media at up to about 1.5 Mbit / s' , Internatiional Standard, 1993 ". However, in this specification, the order of image coding is described normally, but there is no description about the situation when an error occurs. For the detailed coding sequence, refer to the above specifications. In addition, regarding the countermeasures when the error occurs, the Japanese special paper & standard is known to be applicable to the Chinese National Standard (CNS) 84 specifications (210X297 mm) -4-~~ (please read the note on the back first Please fill in this page for more details) 318317 A7 Printed B7 of the Consumer Cooperative of the Central Department of Economic Affairs of the Ministry of Economic Affairs B. V. Description of Invention (2) Ping 6 — 9 8 3 1 3 In the case of coding, when an unavoidable error occurs, the reference image in which the error occurred is replaced by the reference image in the past, so there is a problem of deterioration of the image quality shown later. Here, the reasons for the error during decoding are: (1) There is physical damage to the media holding the CD and other data, and the read data cannot be repaired, that is, when it is damaged; (2) The signal sent by Fu is When the noise caused by thousands of disturbances enters, and the data changes; (3) When encoding, due to the use of methods other than the specifications, the data that would not otherwise appear will be input; etc., such as this case, Decoding the succession becomes impossible, or makes the decoded image very different from the original image. [Problems to be solved by the invention] The data compression of the prediction in the time direction, which is one of the characteristics of the MP EG video specification, is not to transmit or save the image data itself as data in the memory, but to use past and future images and The difference component of the current image is transmitted or stored in the media as data. When decoding, the past or future image is used as the «reference image, and the difference component is used to reproduce the current image. Therefore, it is not limited to the specifications of MPEG video, and uses the prediction in the time direction to achieve the compression technology of the encoded equipment. When the reference image is decoded, data breakage occurs, etc. When the image is damaged, the error will Refer to the image of the following image, or even refer to the image of the image to be broadcast, will (please read the precautions on the back and then fill in this page). The paper size of the binding and binding book is in accordance with Chinese national standards ( CNS) A4 Said Case (210X297mm) -5-Printed by the Central Provincial Bureau of the Ministry of Economic Affairs Employee Consumer Cooperative 318317 A7 ___B7 Fifth, the invention description (3) caused the downward quality of dynamic images. For example, in the MP EG video specification, there are the following types of encoded images. (1) Do not use the predicted image in the time direction. (2) Use the p-type image predicted in the forward direction. ---------- (3) Mutual-type images using bidirectional time prediction. The so-called forward time prediction here means that the actor is expected from the image of the past, and the so-called bidirectional means that the actor is expected from the image of both the past and the future. I-type pictures do not need to refer to pictures during decoding because they do not use time-direction prediction. Since the P-type image uses forward time prediction, the past image is used as the reference image, and the difference component from the reference image is used for decoding. The reference picture here is the type I or P picture that has just been decoded. The B-type image uses the image of both the passing method and the future as a reference image. The referenced here is the I-type or P-type image that is output first, and the I-type or P-type image that is decoded first but output later. Qi 10 is a diagram showing an example of the order of decoding, that is, the order of input images and the order of output images in a decoding device based on MPEG video. In 圚 1 0, 6 1 is the decoded image sequence, 6 1 a ~ 6 In is the decoded images, 6 2 is the output image sequence, and 6 2 a ~ 6 2 m is the The exported image. In addition, the symbols ‘numbers I, P, and B’ added to each image indicate the type of the image, and the number appended to the image type indicates the order in which they are output. In addition, the image column shown in 囵 10 (please read the precautions on the back before filling in this page). Packing and binding The paper size is easy to use China National Standards (CNS) A4 specification (210X297 mm) -6 -Printed and printed A7 __B7_ by the Consumer Cooperative of the Central Ministry of Economic Affairs of the Ministry of Economy V. Description of the invention (4) is an example, it is not necessary to input various types of image data in the order shown in FIG. 10, such as action comparison When there are few animation images, there are even those who have the B-type image data. Conversely, when the motion is more intense, the I-type image data is also input by the consecutive results. In this column, the reference 6 1 a (I 1) that does not need the reference picture is decoded first. Next, using the picture 6 1 a as a reference picture, the picture 6 1 b (P 4) is decoded. Next, using images 6 1 a and 6 1 b as reference images, and images 6 1 c, and similarly using images 6 1 a and 6 1 b as reference images, refer to 6 1 d (B2, B3). Decode sequentially. Next, using picture 6 1 b (P4) as a reference picture, picture 6 1 e (P 7) is decoded. On the other hand, in this example, after decoding of the image 6 1 c (B 2) is started, output starts. The image 6 2 a is the initial state and is not an intentional image. Here, the image 61b (P4) is decoded before the image 61c, 6 Id (B2, B3), but the output is after the image 61c, 61d (B2, B3). When decoding in this order, for example, when decoding an image 6 1 a (II), if an error occurs and is erroneously decoded, use the image 6 1 a (I 1) as the reference image 6 1 b, 6 1 c, 6 Id (P4, B2, B3) also have errors, so the error will be communicated to the image 61e (P7) using image 6 1 b (P 4) as a reference image. In this way, the I-type image 61k (Im) that does not require a reference image appears, and more until only the reference 61n (Pm + 3) using the image 61k (Im) as the reference image is decoded, please read back 1¾ Precautions for refilling this page, binding line, paper size, use Chinese National Standard (CNS) A4 specification (210X297mm) -7-318317 A7 B7 V. Description of invention (5) will be affected by mistakes, resulting in results The quality of the animation is low. The purpose of the present invention is to provide an animation decoding technology that eliminates the influence of errors when any errors are detected during encoding in the encoding method of the animation using the prediction in the time direction, and can avoid the degradation of the image quality. The above and other objects and novel features of the present invention can be understood from the description and drawings in this specification. [Means for Solving the Problem] The above object of the present invention is to investigate the type of the image being decoded at the time when the error is detected during encoding, and it is the type I or P used as the reference circle image until No need to refer to the I-type image of the image until it appears again, skip reading the input data, and after decoding the I-type image that appears, and then until the I-type or P-type image appears, then skip The algorithm for reading input data, the present invention achieves its purpose by adopting this algorithm. 〔Function〕 Printed by the Employee Consumer Cooperatives of the Ministry of Economic Affairs in China (Please read the precautions on the back before filling this page)
I 藉由使用本發明之演算法，即使於解碼中有錯誤被檢 測出時，跳過讀取資料，直到不需要作爲參照圖像之I型 的圖像資料出現爲止，對於之後的圖像資料爲了使解碼再 開，能不使畫質之品質降低，而進行解碼。 〔實施例〕 以下，利用圚式說明本發明之實施例。 本紙張尺度適用中國國家標準（CNS)A4规格（ 210X297公釐）-8 - 經濟部中央樣準局員工消费合作社印裝 A7 B7五、發明説明（6 ) 圖1係將被編碼之動畫予以解碼之動畫解碼裝置之例 ，比例係使用依據MPEG視頻之規格之MPEG視頻解 碼器1 0之動畫再生系統，又圖2係表示該MPEG視頻 解碼器1 0之一實施例。沒有特別的限制，在圖2中，被 1點鎖線4所圍住之各電路方塊，係例如於矽之1個或複 數之半導體基板上所形成，被IC化者。 動畫再生系統，係如圖1所示，由將壓縮、編碼於 C D — R 0M等之蓄積媒體上並被記憶之動畫讀出之驅動 器1 ;及將所讀出之資料分離成圖像資料及聲音資，同時 將控制資訊解讀之微毽腦2;及將被分離之聲音資料解碼 之MPEG視頻解碼器3;及將被分離之動畫資料解碼之 MPEG視頻解碼器1Q所構成。被解碼之動畫，分別被 傳至C R T顯示裝置等之監視器，或被解碼之聲音信號被 傳送至聲頻擴大器。 又，4係上述微電腦2所實行.之程式或固定資料被存 儲之讀出專用記憶體，5係提供微電腦2之作業領業之隨 時讀取寫入可能記憶體，2 0係將上述MPEG視頻解碼 器1 0所連接之被解碼圖像資料，予以暫時保存之緩衝記 憶體。 其次，使用圖2說明MPEG視頻解碼器10的一實 施例。在圖2中，1 1係可將任意的長度（位元數）之圖 像資料進行解碼之可變長解碼器；12係對於被壓縮成頻 率成分之圖像資料，進行反餘弦變換，回復至原來的圖像 資料之反餘弦變換電路；13係由I型之圖像資料及P型 本紙張尺度適用中國國家標準（CNS ) A4規格（210X297公釐）-9 _ (請先閱讀背面之注意事項再填寫本頁) -裝- 訂 A7 B7 經濟部中央榡準局員工消费合作社印裝 五、發明説明（7 ) 或B型之圖像資料復原圖像之動作補償電路；14係將被 復元之圖像資料，存儲於外部之緩衝記憶體2 0 ，而配合 需要讀出之緩衝控制電路；15係依據被復原之圖像資料 ’生成输出RGB信號之影像信號之輸出電路；16係具 有時鐘產生盔1 7，將上述各電路連續控制之複數的控制 信號01 、¢2 、¢3 、04 、05輸出，依據一定之演 算法使圖像資料解碼之內部控制電路。又，上述緩衝記憶 體2 0 ，係可使用例如常用之DRAM構成者，但是也可 以將此緩衝記憶體2 0取入MPEG視頻解碼器內，使全 體成爲1晶片化。上述緩衝記憶體2 0上蓄積有被解碼之 圖像賫料。 在此實施例之動畫解碼裝置（1 〇 )中，可變長解碼 器1 1接受規格外的圖像資料時，形成錯誤檢出信號 ERR，輸出至上述控制電路1 6 ，控制電路1 6接收至 此錯誤檢出信號E R R，則進行以下所述之復原處理而構 成者。 其次，關於上述動畫裝置1 〇之解碼錯誤發生時之復 原處理，使用圖3之流程圖說明。復原處理係隨著解碼裝 e之電路構成，方式等而不同，所以本實施例僅只係表示 —例者。 說明圖3之流程圖之流程時，係如下述。 通常之解碼動作中，發生錯誤時，控制會移轉至圖3 之狀態1 0 1。然後，在狀態1 〇 2，錯誤發生之圖像之 型可成爲參照圖像之I型或P型，或判斷是否不成·爲參照 κ士 I 1 I (讀先閲讀背面之注意事項再填寫本頁) 線 本紙張尺度適用中國國家標準（CNS)A4規格（210X297公釐）—1〇 經濟部中夬標準局員工消費合作社印製 318317 A7 _B7五、發明説明（8 ) 圚像之B型，隨著其結果，分別分岐至各程序。又，各圖 像資料之型係依據被输入之圖像資料之先頭所付記之被稱 爲標題的部份所存儲之資訊來判定。 此處，解碼之錯誤所發生之圖像之型可成爲參照圖像 之I型或P型時，將該圚像作爲參照圖像使用之圖像，爲 了接收錯誤的影響，使被解碼之結果不能成爲正確者。更 者，以接受錯誤的影響之圖像作爲參照圖像之圖像也會受 到錯誤之影響。所以於I型或P型的圖像，發生解碼的錯 誤時，從不需要參照圖像之I型之圇像再度出現的時刻起 ，不使解碼再開，則無法排除錯誤的影響。 於是在此實施例，發生錯誤之圖像爲I型或P型時， 控制被移至狀態1 0 3 ，在此狀態，以檢索下一個圖像資 料的先頭.（標題），跳過讀取發生錯誤之圖像之資料，於 下一個狀態之先頭，判斷圇像之先頭是否被檢測出。然後 ，當圖像資料之先頭沒被檢出時，重覆狀態1 0 4直到被 檢測出爲止。 於狀態1 0 4 ，當下一個圇像資料之先頭被檢測出時 ，檢移至狀態1 〇 5，從被稱爲圖像開始符號之資料的先 頭資訊之中，讀出該圖像之型，進行判斷該圖像是否爲I 型，從前述之理由，當爲I型以外的圖像時，控制回到狀 態1 0 3，再度進行下一個圖像之檢索。於狀態1 0 5， 當I型之圖像被檢測出時，於狀態1 0 6該圖像被解碼。 此處，爲了使B型之圖像正確地被解碼，必需要2個 正確地被解碼之I型或P型之圖像。所以，無法立即從狀 a ( CNS ) Α4ί)Μί· ( 210X297^* ) - 11 -— (請先閱讀背面之注意事項再填寫本頁) / -裝 訂 線 經濟部中央標準局員工消費合作社印裝 A7 B7五、發明説明（9 ) 態1 0 6立刻回復至通常之解碼順序，在解碼終了，控制 移至下一個狀態1 0 7，在此狀態，與狀態1 0 3相同， 進行下一個圖像之先前檢索。然後，當下一個圖像的先頭 被檢出時，於狀態1 0 8進行圖像之型的判別，當爲不能 作爲參照圖像使用之B型時，控制被轉移至狀態1 0 9 ， 下一個圖像之先頭的檢索被進行，然後控制再回到狀態 10 7° 於狀態1 0 8，可作爲參照圖像所使用之I型或P型 之圖像被檢出時，爲了使以後依照通常的解碼順序之控制 被移至狀態1 1 2 ，除了被保持於緩衝記憶體2 0之顯示 中的圖像資料之、外，其他的圖像資料都作廢，之後回復至 通常的解碼順序。又，此狀態112也可以不要在回復前 進行而在進入錯誤處理之後進行。 另一方面，在流程圖的先頭之狀態1 0 2處，造成錯 誤之圖像被判斷爲B型時，由於B型圇像不作爲參照圖像 使用，所以必需作廢的是只有該圖像。所以，轉移到狀態 1 1 〇，以進行下一個圖像資料之檢索，於跳過讀取錯誤 發生之資料後，於狀態1 1 1處，判定下一個圖像之先頭 是否檢測出，在檢測出的時刻，回復到通常之解碼順序。 圖5係表示發生錯誤之圖像爲B型之圖像時之输出時 序，而圖6係表示發生錯誤之圖像爲P型之圖像時之輸出 時序。 如圖5所示，造成錯誤之圖像爲B型（例如B 3 )時 ，只跳讀該圖像（B3 )而其1個前面的圖像（B2 )被 本紙張尺度適用中國國家橾準（CNS)A4規格（210父2们公釐）-12 - (請先Μ讀背面之注意事項再填寫本頁) .裝. 訂 線 A7 B7 經濟部中夬標準局員工消費合作社印製 五、發明説明（10) 繼縝输出。另一方面，如圖6所示，造成錯誤的圖像爲P 型時（例如Pm+4)，以該圖像及這個圖像作爲參照圖 像之圓像資料（Bm+l，Bm+2)，當然在這個之後 ，直到下一個I型之圖像（I η )出現爲止，所有的圖像 都要放棄，其間，之後的输出圖像（Ρπι )繼績被输出。 圖4係表示本發明之動畫解碼裝置之解碼錯誤發生時 之復原處理之其他的例子之流程圚。 在圖3之實施例中，對應解碼錯誤所發生之圖像的型 ，進行非常細地補救處理，但是要以積體電路上的邏輯電 路等實現此復原處理時，只因爲是多機能，所以導致電路 規模之增大，而且會造成成本之增加。於是本實施例是採 用以削減機能縮減電路規模爲目的之方式。 本流程圖之流程說明如下。 當解碼錯誤被檢測出時，控制被傳至狀態2 0 1 圖3之實施例，此處進行圚像之型的判別，但是在本實施 例中，於下一個狀態2 0 2中，以進行下一個圖像資料之 先頭檢索，使跳過讀取發生錯誤之資料之處理被進行’於 狀態2 0 3中判定圚像資料之先頭是否被檢測出。 當下一個圚像資料被檢測出時，於下一個狀態2 〇 4 中，進行該圚像之型的識別，如果不是I型之圖像，再使 控制回到狀態2 0 2。於狀態2 0 4中，當I型之圖像被 檢測出時，使控制轉移至狀態2 0 5，使該圖像解碼。 解碼終了後，爲了不使下一個出現之Β型之圖像解碼 ，於狀態2 0 6檢測出圚像之先頭，於狀態2 0 7判斷該 (請先閲讀背面之注意事項再填寫本頁) 裝- 訂 線 本紙張尺度適用中國國家榡準（CNS)A4说格（ 210X297公釐）-13- A7 B7 經濟部中央標準局員工消费合作社印製 五、發明説明（11 ) 圖像是否爲B型，如果是B型轉移至狀態2 0 8 ，重覆下 一個圖像之檢索直到成爲B型以外爲止，在成爲B型以外 之時刻，使控制轉移至狀態2 0 9，使除了保持於緩衝記 憶體之錯誤檢測前之顯示中圖像資料之外，其他的圖像資 料全部放棄，回復至通常之解碼動作。 藉由使用以上之流程圚，只要作廢原來的發生錯誤之 圖像，即使是良好的B型之圖像的錯誤時也會多少發生圖 像之缺落，但是以更簡單之電路構成，演算法可實現從解 碼錯誤起之回復處理。 圖7係表示在圖4的實施例中，在B型之圖像發生解 碼錯誤時之輸出圖像。在圖7中，3 1爲被解碼之輸入圖 像列，而從3 1 a至3 1 d及從3 1 i至3 1 k係被輸入 資料之中被解碼者，從31e至31h係被输入資料之中 ，被跳過讀取者。3 2係输出圖像列。在此實施例，於不 能成爲參照圖像之B型的圖像處，產生錯誤時，在該時刻 ，所有被存儲於緩衝記憶體之圖像資料（含有正常之圖像 Pn + 3 )被作廢，而且直到下一個I型之圖像（I η ) 出現爲止，所有的圖像資料被跳過讀取。將圖5與圖7相 比較，則可以了瞭圖3之實施例與圖4之實施例之處理的 方法之不同。 另一方面，不使錯誤所發生之圖像作棄時，如圖9所 示，發生錯誤，藉由输出解碼失敗之圖像（Bm+2)， 導致輸出的品質降低。圖9中，5 1係被解碼之輸入圖像 列，從5 1 a至5 Id及從5 1 i至5 lk，係被輸入之 (請先閱讀背面之注意事項再填寫本頁) -'9 本紙張尺度逍用中國國家標準（CNS)A4规格（2!0X297公釐）-14 - _ B7 五、發明説明（12 ) 資料之中被解碼者，而5 1 e至5 i h係被輸入資料之中 ，被跳過讀取者。5 1爲輸出圚像列，5 2 e係具有使發 生錯誤之輸入圖像51d (Bm+2)復合之缺陷之输出 圚像。 兩圚所共通之圖中記號，係表示圖像之型及输出順序 。例如，I η — 1係表示爲I型之圖像，被輸入至第^ — 1個。 在圖9中，於圖像5 1 d (Bm+2 )之解碼中，由 於發生錯誤，所作爲從錯誤來的復原動作，跳過讀取 5 1 e至5 1 h之輸入資料，其次從出現之I型之圖像之 圖像51i起，使解碼再開，回復到通常的動作。 經濟部中夬棵準局員工消費合作社印裝 (請先閏讀背面之注意事項再填寫本頁) 此處，於錯誤被檢測出之時刻之輸出之圖像，在此例 中係爲圖像5 2 d (Bm+1 )，作爲未輸出的圖像而言 ，直圖像5 1 d (Pm+3 )，發生錯誤之圚像5 1 d ( Bm+2)，但是在錯誤來的復原處理終了爲止，都持績 地輸出5 2d (Bm+1)。此處，當發生錯誤時，分配 至圖像5 1 d (Bm+2 )之圖像記憶體先不作廢，則錯 誤之復原後圖像52e (Bm+2) 、52f (Pm+3 )之顯示被持繼，而圖像5 2 e (Bm+2 )因爲發生錯 誤，所以解碼不完全，如此則會導致動畫品質之低下。 對於此，依照圖4之實施例，則如圖7所示之時序程 序，錯誤發生前後之圖像被作廢，藉由錯誤來的復原，從 解碼被再開的部份起，以新穎之順序，使输出被>|開，不 、會有输出具有缺陷之圖像，而可以使動畫品質的降低控制 本紙張尺度適用中國國家橾準（CNS)A4規格（210X297公釐）_ 15 - 318317 A7 B7 經濟部中央搮準局員工消費合作社印製 五、發明説明（13) · 在最小。 在圖7中，到錯誤發生爲止之處理，係與圖9相同， 但是在錯誤發生的時刻，將除了输出中的圖像（在本例爲 圖像3 2 d )以外之圇像作廢，錯誤來的復原完了，在輸 出之繼績爲可能之時刻，作爲將I型之圖像31 i ( I η )解碼之圖像3 2 e，重新輸出。 又，如圖8所示，也可以將作廢之圖像限定在有缺陷 之圖像。圖8中，4 1係被解碼之输入圖像到，4 1 a至 4 1 d及4 1 i至4 1 k，係被输入之資料之中被解碼者 ，4 1 e至4 1 h係被輸入資料之中，被跳過讀取者。 4 2係输出圖像列。在此實施例中，於錯誤發生時被作廢 之圖像，只有發生錯誤之圖像4 1 d (Bm+2 )，於發 生錯誤時，解碼的終了之正常圖像4 1 b (Pm+3 )被 作廢，從錯誤復原時，作爲圖像4 2 e而被輸出。然後， 之後，藉由再開之解碼，使被生成之圖像繼續輸出。藉此 ，使動晝的圆滑性受到補償。 又，於解碼處理中，當錯誤被檢測出時，發生錯誤的 不是現在解碼中的圖像，而係一個之前之解碼中，參照圓 像之解碼的階段所發生之可能性很大。所以，雖然錯誤被 檢測出，但是當錯誤的發生時刻之特定很困難時，將有可 能之圚像全部作廢，希望能控制畫質的品質之降低。 另一方面，於圖3及圖4之實施例中，也可以使保持 於緩衝記憶體之未輸出之圖像作廢之步驟1 1 2 ，2 0 9 予以省略。像這樣作，至少使檢測出錯誤之圖像及其後參 (請先閱讀背面之注意事項再填寫本頁) 、-° Γ 本紙張尺度適用中國國家標準（CNS )八4規格（2!0X297公釐）-16 - 經濟部中央標準局員工消費合作社印裝 A7 B7 五、發明説明（14) 照圖像出現爲止之输入圖像被跳過讀取而被作廢，所以可 以畫質的低下控制在某一程度之下。又，在錯誤被檢測出 的時刻，與將緩衝記憶體所保持之未输出之圖像全部作廢 之方法相比較，可使動作的園滑性得到某種程度之補償。 如以上之說明，上述實施例，於將解碼之圖像作爲參 照圖像使用時之動畫解碼裝置中，解碼時有任何之錯誤被 檢測出且以錯誤被檢測出之圖像作爲參照圇像使用之圖像 時，以可作爲參照圖像使用之圖像，使輸入圖像作廢直到 不受錯誤之影響而解碼可能之圖像出現爲止，使該圖像解 碼後，直到參照參照圖像之圖像出現爲止，使輸入圖像作 廢，在使該圖像解碼後，使通常之解碼再開，所以於解碼 中，即使當錯誤被檢測出時，直到不需要參照圇像之型的 圖像資料出現爲止，使输入資料被跳過讀取，對於之後的 圚像，由於解碼再開，所以具有不會使畫質降低，而能進 行解碼之效果。 又，於解碼時有任何的錯誤被檢測出之圖像時，於作 爲參照圚像使用之圖像，直到不受錯誤的影響而使解碼成 爲可能之圖像出現爲止，使输入圖像作廢，使該圖像解碼 後，直到參照該參照圖像之圖像出現爲止，使輸入盡像作 廢，由於將該圖像解碼後，使通常的解碼再開，所以不使 用複雜之硬體，而具有可以進行使塞質品質降低情形減少 之解碼之效果。 更者，於解碼時，當錯誤被檢測出時，使解碼結束且 使未输出之圖像全部作廢，直到通常之解碼再開爲止，於 本紙張尺度適用中國國家標準（CNS ) A4规格（210X297公釐）-17 - _ ' (請先閲讀背面之注意事項再填寫本頁) .裝 訂 A7 B7 五、發明説明（15) 錯誤被檢測出時刻，重覆輸出被輸出之物，所以可以使構 成更爲簡單。 以上已依據本發明之具體實施例，作了說詳細說明， 但是本發明並不只限於上述之實施例，只要在其不離開本 要旨之範圍內，可作種種的變更。例如在上述實施例中， 已經對以動畫解碼裝置作爲專用之半導體稹體電路而構成 時之情況作了說明，但是也適用於使用汎用之D S P來構 成者。這時，本發明係藉由D S P之動作程式即軟體來實 現。 以上之說明，主要係以發明背景之利用領域之 MP EG視頻解碼器爲例作了說明，但是只要是使用時間 方向之預測方式之動晝解碼裝置，都可適用本發明。 〔發明之效果〕 將本發明之效果作一簡單之說明。 如以上之說明，本發明採用簡單三演算法，於動畫解 碼裝置中，於解碼錯誤發生時，可使畫質之惡化控制在最小。 〔圖式說明〕 圖1係表本發明之動晝解碼裝置之一實施例之動盡再 生統之構成係之方塊圖。 ， 圖2係表示MP E G視頻解碼器之一實施例之方塊圖 (請先閲讀背面之注意事項再填寫本頁)I By using the algorithm of the present invention, even when an error is detected in decoding, the reading of data is skipped until the type I image data that does not need to be a reference image appears. For subsequent image data In order to restart decoding, decoding can be performed without degrading the quality of the picture quality. [Embodiments] Hereinafter, embodiments of the present invention will be described using a formula. This paper scale is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) -8-Printed by the Central Sample Bureau of the Ministry of Economic Affairs Employee Consumer Cooperative A7 B7 V. Invention description (6) Figure 1 decodes the encoded animation For an example of an animation decoding device, the ratio is based on the animation reproduction system of the MPEG video decoder 10 according to the MPEG video specifications. FIG. 2 shows an embodiment of the MPEG video decoder 10. There is no particular limitation. In FIG. 2, each circuit block surrounded by the 1-point lock wire 4 is formed on one or a plurality of semiconductor substrates of silicon, for example, and is ICized. The animation reproduction system, as shown in Figure 1, consists of a drive 1 that compresses, encodes on a storage medium such as CD-ROM and reads out the animation; and separates the read data into image data and Audio data, at the same time, the micro-shuttle brain 2 which controls the interpretation of information; and the MPEG video decoder 3 which decodes the separated audio data; and the MPEG video decoder 1Q which decodes the separated animation data. The decoded animations are sent to monitors such as CRT display devices, or the decoded sound signals are sent to audio amplifiers. In addition, 4 is the dedicated memory for reading the program or fixed data stored in the above microcomputer 2. 5 is the memory for reading and writing at any time for the operation of the microcomputer 2. The 20 is the above MPEG video. The buffered memory of the decoded image data connected to the decoder 10 is temporarily saved. Next, an embodiment of the MPEG video decoder 10 will be described using FIG. In Figure 2, 1 1 is a variable-length decoder that can decode image data of arbitrary length (number of bits); 12 is an inverse cosine transform of the image data compressed into frequency components, reply Inverse cosine transform circuit to the original image data; 13 is composed of I-type image data and P-type paper. The paper standard is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) -9 _ (please read the back (Notes and then fill out this page)-Installation-Order A7 B7 Printed and printed by the Central Consumer ’s Consumer Cooperative of the Ministry of Economic Affairs V. Invention description (7) or type B image data. Motion compensation circuit for image restoration; 14 series will be The restored image data is stored in an external buffer memory 2 0, and cooperates with the buffer control circuit that needs to be read; 15 is an output circuit that generates an image signal to output RGB signals based on the restored image data; 16 is The clock generates a helmet 17 and outputs a plurality of control signals 01, ¢ 2, ¢ 3, 04, 05 that are continuously controlled by the above circuits, and an internal control circuit that decodes the image data according to a certain algorithm. In addition, the buffer memory 20 described above can be constructed using, for example, a commonly used DRAM, but the buffer memory 20 can also be taken into an MPEG video decoder to make the entire chip into one chip. The decoded image data is stored in the buffer memory 20. In the animation decoding device (10) of this embodiment, when the variable-length decoder 11 receives out-of-specification image data, an error detection signal ERR is formed, which is output to the above-mentioned control circuit 16 and the control circuit 16 receives So far, the error detection signal ERR is constructed by the following restoration process. Next, the recovery process when the decoding error of the animation device 10 occurs will be described using the flowchart of FIG. 3. The restoration process differs according to the circuit configuration, method, etc. of the decoding device, so this embodiment is only shown as an example. The flow of the flowchart of FIG. 3 is explained as follows. In the normal decoding operation, when an error occurs, the control will shift to the state 101 in FIG. 3. Then, in state 1 〇2, the type of the image where the error occurred can become the type I or P of the reference image, or judge whether it is not possible. For the reference κ 士 I 1 I (read the precautions on the back before filling in this Page) The size of the line paper is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm)-10 Printed 318317 A7 _B7 by the employee consumer cooperative of the Zhongshang Standards Bureau of the Ministry of Economy. Along with the result, they are divided into different procedures. In addition, the type of each image data is determined based on the information stored in the part called the title that is paid at the beginning of the input image data. Here, when the type of the picture where the decoding error occurs can be the I-type or P-type of the reference picture, use the image as the reference picture, in order to receive the influence of the error, the result of the decoding Can't be the right one. Furthermore, images that use the image affected by the error as a reference image will also be affected by the error. Therefore, when a decoding error occurs in an I-type or P-type image, from the time when the I-type image of the reference image does not need to appear again, the effect of the error cannot be ruled out without restarting the decoding. So in this embodiment, when the error image is I type or P type, the control is moved to state 1 0 3, in this state, to retrieve the head of the next image data. (Title), skip reading At the beginning of the next state, the data of the image where the error occurred is judged whether the beginning of the image is detected. Then, when the head of the image data is not detected, the status 104 is repeated until it is detected. In state 1 0 4, when the head of the next image data is detected, it moves to state 1 05, and the type of the image is read from the head information of the data called the start symbol of the image. It is judged whether the image is of type I. For the reasons mentioned above, when it is an image other than type I, control returns to state 103, and the next image search is performed again. In state 105, when an image of type I is detected, the image is decoded in state 106. Here, in order to correctly decode the B-type image, two I-type or P-type images must be correctly decoded. Therefore, it is not possible to immediately print out the status a (CNS) Α4ί) Μί · (210X297 ^ *)-11 -— (please read the precautions on the back and then fill out this page) /-binding line, the Ministry of Economy Central Standards Bureau employee consumer cooperative printing A7 B7 V. Description of the invention (9) State 1 0 6 immediately reverts to the normal decoding order. At the end of decoding, the control moves to the next state 1 0 7, in this state, the same as state 1 0 3, proceed to the next figure Like the previous search. Then, when the head of the next image is detected, the image type is judged in state 108, and when it is a type B that cannot be used as a reference image, control is transferred to state 1 0 9, the next The first search of the image is carried out, and then the control returns to the state 10 7 °. In the state 1 0 8, when the type I or P type image that can be used as the reference image is detected, in order to make the future follow the normal The control of the decoding order is moved to state 1 1 2. Except for the image data held in the display of the buffer memory 20, all other image data are invalidated, and then the normal decoding order is restored. In addition, this state 112 may not be performed before the reply but after entering the error processing. On the other hand, when the error-causing image is judged to be type B at the first state 102 in the flowchart, since the type B image is not used as a reference image, it is necessary to invalidate only this image. So, move to state 1 1 〇 to retrieve the next image data. After skipping the data where the reading error occurred, at state 11 1 it is determined whether the head of the next image is detected. At the outgoing moment, return to the normal decoding order. Fig. 5 shows the output timing when the error image is a B-type image, and Fig. 6 shows the output timing when the error image is a P-type image. As shown in Figure 5, when the image that caused the error is type B (for example, B 3), only the image (B3) is skipped and its previous image (B2) is approved by the Chinese standard for this paper. (CNS) A4 specification (210 parents 2 mm) -12-(please read the precautions on the back and then fill out this page). Packing. Thread A7 B7 Printed by the Employee Consumer Cooperative of the Bureau of Standards and Statistics, Ministry of Economic Affairs Description of the invention (10) Following the meticulous output. On the other hand, as shown in Figure 6, when the image causing the error is P-type (for example, Pm + 4), the image and this image are used as the reference image for the round image data (Bm + 1, Bm + 2 ), Of course, after this, until the next I-type image (I η) appears, all images must be abandoned, during which the subsequent output image (Pπι) successor is output. Fig. 4 is a flowchart showing another example of recovery processing when a decoding error occurs in the video decoding device of the present invention. In the embodiment of FIG. 3, very fine remedial processing is performed according to the type of the image where the decoding error occurs, but when the restoration processing is to be implemented by a logic circuit on an integrated circuit, etc., it is only because it is multi-functional. This leads to an increase in circuit scale and an increase in cost. Therefore, this embodiment adopts a method for reducing the function and reducing the circuit scale. The flow of this flowchart is explained as follows. When a decoding error is detected, control is transferred to state 2 0 1 in the embodiment of FIG. 3, where the image type is determined, but in this embodiment, in the next state 2 0 2 to perform The first search of the next image data, so that the processing of skipping the reading of the data with errors is carried out 'in state 2 0 3 to determine whether the head of the image data is detected. When the next image data is detected, in the next state 204, the type of the image is identified. If it is not an I-type image, control is returned to state 202. In state 204, when an I-type image is detected, control is transferred to state 205 to decode the image. After decoding, in order not to decode the next type B image, the head of the image is detected in state 206, and it is judged in state 207 (please read the precautions on the back before filling this page) Binding-The paper size of the binding book is applicable to the Chinese National Standard (CNS) A4 (210X297mm) -13- A7 B7 Printed by the Employees ’Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs V. Invention Instructions (11) Is the image B Type, if it is type B, transition to state 2 0 8, repeat the search of the next image until it becomes other than type B, and at the time when it becomes other than type B, the control is transferred to state 2 0 9, so that it remains in the buffer Except for the image data in the display before the error detection of the memory, all other image data are discarded and the normal decoding action is restored. By using the above process, as long as the original error image is discarded, even a good B-type image error will cause some image defects, but with a simpler circuit configuration, the algorithm It can realize the reply processing from decoding error. Fig. 7 shows the output image in the embodiment of Fig. 4 when a decoding error occurs in the B-type image. In Fig. 7, 3 1 is the input image row to be decoded, and from 3 1 a to 3 1 d and from 3 1 i to 3 1 k are decoded among the input data, from 31e to 31h. In the input data, the reader is skipped. 3 2 series output image column. In this embodiment, when an error occurs in a B-type image that cannot be a reference image, at that moment, all image data stored in the buffer memory (including the normal image Pn + 3) is invalidated , And until the next I-type image (I η) appears, all the image data is skipped for reading. Comparing FIG. 5 with FIG. 7, it can be seen that the processing methods of the embodiment of FIG. 3 and the embodiment of FIG. 4 are different. On the other hand, when the image where the error occurs is not discarded, as shown in Fig. 9, an error occurs, and the image that failed to be decoded is output (Bm + 2), which leads to a decrease in the quality of the output. In Figure 9, 5 1 is the input image column to be decoded, from 5 1 a to 5 Id and from 5 1 i to 5 lk, it is input (please read the precautions on the back before filling this page)-' 9 This paper standard uses the Chinese National Standard (CNS) A4 specification (2! 0X297 mm) -14-_ B7 5. Description of the invention (12) The decoded person in the data, and 5 1 e to 5 ih are input The reader was skipped among the materials. 5 1 is the output image sequence, and 5 2 e is the output image with the defect that the input image 51d (Bm + 2) that caused the error is compounded. The marks in the picture shared by the two schools indicate the image type and output order. For example, I η — 1 is represented as an I-type image and is input to the ^ — 1st. In FIG. 9, in the decoding of the image 5 1 d (Bm + 2), due to an error, the recovery action from the error skips reading the input data from 5 1 e to 5 1 h, followed by When the image 51i of the type I image appears, the decoding is restarted and the normal operation is restored. Printed by the Ministry of Economic Affairs of the Central Bureau of Precinct Employee Consumer Cooperative (please read the precautions on the back before filling this page) Here, the output image at the time when the error is detected, in this case it is an image 5 2 d (Bm + 1), as a non-output image, the straight image 5 1 d (Pm + 3), the error image 5 1 d (Bm + 2), but recovered from the error Until the end of the process, 5 2d (Bm + 1) is output with success. Here, when an error occurs, the image memory allocated to image 5 1 d (Bm + 2) will not be invalidated first, then the image 52e (Bm + 2) and 52f (Pm + 3) after the error recovery The display is continued, and the image 5 2 e (Bm + 2) is decoded incompletely due to an error, which will result in poor animation quality. Regarding this, according to the embodiment of FIG. 4, in the sequence procedure shown in FIG. 7, the images before and after the error are discarded, and the recovery from the error is performed in a novel order from the part where the decoding is reopened. Enable the output to be turned on. No, there will be output of defective images, and the quality of the animation can be reduced. This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) _ 15-318317 A7 B7 Printed by the Employee Consumer Cooperative of the Central Department of Economics of the Ministry of Economy V. Description of Invention (13) · At the smallest. In FIG. 7, the processing until the error occurs is the same as in FIG. 9, but at the time of the error, the images other than the output image (image 3 2d in this example) are discarded and the error occurs. After the restoration is completed, when the output succession is possible, the image 3 2 e is decoded as the I-type image 31 i (I η) and output again. Also, as shown in Fig. 8, the invalid image may be limited to a defective image. In Fig. 8, 4 1 is the input image to be decoded, 4 1 a to 4 1 d and 4 1 i to 4 1 k are the decoded among the input data, 4 1 e to 4 1 h The reader is skipped in the input data. 4 2 series output image column. In this embodiment, only the image 4 1 d (Bm + 2) that was discarded when the error occurred, and the normal image 4 1 b (Pm + 3) that was decoded when the error occurred ) Is invalidated and output as an image 4 2 e when recovering from an error. Then, afterwards, by reopening the decoding, the generated image continues to be output. With this, the smoothness of the moving day is compensated. Also, in the decoding process, when an error is detected, it is not the image currently being decoded but the previous decoding, which is likely to occur at the stage of decoding with reference to the circular image. Therefore, although an error is detected, when it is difficult to specify the timing of the error, all possible images will be discarded, and it is hoped that the quality of the image will be reduced. On the other hand, in the embodiments of FIGS. 3 and 4, the steps 1 1 2 and 2 0 9 of discarding the unoutput images held in the buffer memory may be omitted. To do so, at least make the wrong image and its reference (please read the precautions on the back before filling in this page),-° Γ This paper size is applicable to the Chinese National Standard (CNS) 84 specifications (2! 0X297 Mm) -16-A7 B7 printed by the Employees ’Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economy V. Description of the invention (14) The input image is skipped for reading until the image appears, so the image quality can be reduced Below a certain level. In addition, when the error is detected, compared with the method of discarding all unoutput images held in the buffer memory, the smoothness of the operation can be compensated to some extent. As explained above, in the above embodiment, in the animation decoding device when the decoded image is used as the reference image, any error is detected during decoding and the image with the error detected is used as the reference image In the case of an image, use an image that can be used as a reference image to invalidate the input image until an image that can be decoded without being affected by the error appears. After decoding the image, refer to the image of the reference image Before the image appears, the input image is invalidated, and after decoding the image, the normal decoding is restarted, so during decoding, even when an error is detected, until the image data of the reference image type does not need to appear So far, the input data has been skipped, and the decoding of the next image is reopened, so there is an effect that decoding can be performed without degrading the image quality. In addition, when there is any error detected image during decoding, the image used as the reference image will be invalidated until the image that is not affected by the error makes decoding possible. After the image is decoded, until the image referring to the reference image appears, the input is discarded as much as possible. Since the image is decoded and the normal decoding is reopened, it does not use complicated hardware, but can The effect of decoding that reduces the quality of the plug is reduced. Furthermore, when decoding, when an error is detected, the decoding is completed and all unoutput images are discarded until the normal decoding is reopened. The Chinese National Standard (CNS) A4 specification (210X297 %)-17-_ (Please read the precautions on the back before filling in this page). Binding A7 B7 V. Description of invention (15) When the error is detected, the output is repeated, so the composition can be made more For simplicity. The above has been described in detail based on the specific embodiments of the present invention, but the present invention is not limited to the above-mentioned embodiments, as long as it does not depart from the scope of the gist, various changes can be made. For example, in the above-mentioned embodiment, the case where the video decoding device is configured as a dedicated semiconductor block circuit has been described, but it is also applicable to those who use a general-purpose DSSP. At this time, the present invention is realized by the operation program or software of DSSP. The above description is mainly based on the MP EG video decoder used in the background of the invention as an example. However, the present invention can be applied to any dynamic day decoding device that uses a prediction method in the time direction. [Effects of the Invention] The effects of the present invention will be briefly explained. As explained above, the present invention uses a simple three-pass algorithm to minimize the deterioration of image quality when decoding errors occur in the animation decoding device. [Description of the drawings] FIG. 1 is a block diagram showing the constitution of the motion reproduction system of an embodiment of the motion day decoding device of the present invention. , Figure 2 is a block diagram showing an embodiment of an MP E G video decoder (please read the precautions on the back before filling this page)
S 丁 % 本紙張尺度適用中國國家標準（CNS)A4規格（ 210X297公釐）_ 18 - 318317 A7 B7 五、發明説明（16 ) 圈3係表示實施例之動盡解碼裝S之解碼錯誤檢出時 之復原處理之順序之一實施例之流程圇。 圖4係表示實施例之動畫解碼裝置之解碼錯誤檢出時 之復原處理之順序之一實施例之流程圖。 圖5係依據圖3之順序之動晝解碼裝®之B型圖像之 錯誤檢測時之圖像輸出時序圖。 圖6係依據圖3之順序之動畫解碼裝置之P型之圖像 錯誤檢測時之圚像輸出之時序圖。 圓7係依據圖4的順序之動晝解碼裝置之B型圖像之 錯誤檢測時之圖像输出之時序圖。 圖8係其他的實施例之動晝解碼裝置之B型圇像之錯 誤檢出時之圚像輸出之時序圖。 圖9係習知之動畫解碼裝置之錯誤檢出時之圖像輸出 之時序。 圖1 0係表示動畫解碼裝置之例之MPEG視頻解碼 器之解碼及圇像輸出之一實施例之時序圖。 〔記號說明〕 經濟部中央樣準局貝工消费合作社印製 (請先閲讀背面之注意事項再填寫本頁) 10 動晝解碼裝e 2 0 緩衝記憶體 101〜112 流程圖之狀態 201〜209 流程圖之狀態 3 1 解碼中之圖像群 31 a〜31k 各解碼中之圖像 本紙張尺度適用中國國家標準（CNS)A4说格（ 210X297公釐）-19 - 31S317 A7 B7 五、發明説明（17 ) 3 2 輸出中的圖像 32a〜32f 各輸出中之圖像 4 1 解碼中之圖像群 4 la〜4 lk 各解碼中的圖像 4 2 輸出中的圖像 42a〜42f 各输出中之圖像 51 解碼中之圖像群 5 2 输出中之圖像群 52a〜52f 各輸出中之圖像 6 1 解碼中的圇像群 61a〜61η 各解碼中的圖像 6 2 輸出中之圖像 62a〜62η 各輸出中之圖像 (請先閲讀背面之注意事項再填寫本頁) .裝 訂 線 經濟部中夬樣準局員工消費合作杜印裝 本紙張尺度適用中國國家標準（CNS)A4規格（ 210X297公釐）_ 20 -S 丁% This paper scale is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) _ 18-318317 A7 B7 V. Description of the invention (16) Circle 3 represents the decode error detection of the exhaustive decoding device of the embodiment. The sequence of the embodiment of the sequence of the restoration process in time. Fig. 4 is a flowchart showing an embodiment of a sequence of recovery processing when a decoding error is detected by the animation decoding device of the embodiment. Fig. 5 is a timing diagram of the image output during the error detection of the B-type image of the moving day decoding device according to the sequence of Fig. 3. FIG. 6 is a timing diagram of image output during P-type image error detection of the animation decoding device according to the sequence of FIG. 3. Circle 7 is a timing chart of image output during error detection of the B-type image of the moving day decoding device according to the sequence of FIG. 4. Fig. 8 is a timing chart of the output of the image when the error of the B-type image of the dynamic day decoding device of the other embodiment is detected. FIG. 9 is the timing of image output when an error is detected in a conventional animation decoding device. Fig. 10 is a timing chart showing an example of decoding and image output of an MPEG video decoder as an example of a video decoding device. [Description of symbols] Printed by the Beigong Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs (please read the precautions on the back before filling in this page) 10 moving day decoding device e 2 0 buffer memory 101 ~ 112 flow chart status 201 ~ 209 State of the flow chart 3 1 Decoded image group 31 a ~ 31k Each decoded image The paper size is applicable to the Chinese National Standard (CNS) A4 format (210X297 mm) -19-31S317 A7 B7 V. Description of the invention (17) 3 2 Images under output 32a to 32f Images under output 4 1 Images under decoding 4 la to 4 lk Images under decoding 4 2 Images under output 42 a to 42 f Image 51 in progress Image group in decode 5 2 Image group in output 52a to 52f Image in each output 6 1 Image group in decode 61a to 61η Image in decode 6 2 Output Images 62a ~ 62η Images in each output (please read the precautions on the back before filling in this page). Binding Line Ministry of Economy, China Bureau of Standards and Samples, consumer cooperation, du printing, the paper size is applicable to the Chinese National Standard (CNS) A4 size (210X297mm) _ 20-