JP2004064115A - Code amount control apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a code amount control apparatus capable of executing code amount control applied to a bit stream including compression encoding data at high speed with less computational complexity so as to suppress distortion with respect to a rate. <P>SOLUTION: In the code amount control apparatus 1, an MMU 3 temporarily stores an input bit stream including compression encoded data subjected to compression encoding by the JPEG 2000 system into a storage device 2, reads data OD from the storage device 2 according to control signals CS1, CS2 and outputs the data to a multiplexer section 5. The multiplexer section 5 multiplexes the data OD and outputs the multiplexed data as an output bit stream. A bit truncation control section 4 is provided with an image quality control section 10 for selecting an encoding object in conformity with the target image quality and an encoded amount control section 11 for controlling the encoded amount in conformity with the target encoding amount. A layer division control section 7 outputs a read control signal CS2 to generate an output bit stream divided into a plurality of layers to the MMU 3. <P>COPYRIGHT: (C)2004,JPO

Description

【００４８】
上式（１）中、ｚは、ビット切り捨て点（ｂｉｔ　ｔｒｕｎｃａｔｉｏｎ　ｐｏｉｎｔ）；ｏｙ_ｉ ^{Ｋ［ｉ，ｊ］}［ｊ］は、Ｋ［ｉ，ｊ］番目のビットプレーンで逆量子化されたコードブロックのｊ番目のサンプル値（係数値）；ｙ_ｉ［ｊ］は、当該コードブロックのｊ番目のサンプル値（係数値）；Ｇ_ｂ［ｉ］は、サブバンドｂ［ｉ］に対応する合成フィルタ係数のノルムの二乗であって、当該サブバンドｂに依存する歪モデルの重み係数を示している。尚、説明の便宜上、上式（１）に示した記号の表記法は、前記参考文献Ａでのそれとは若干異なる。
【００４９】
レート・歪み最適化では、この歪測度Ｄ_ｉ ^（ｚ）のサブバンドｂ［ｉ］における総和量を最小にするような最適化処理が行われる。サブバンドｂの重み係数Ｇ_ｂは、画像の歪みを低減させるための重み付けを表している。
【００５０】
サブバンドｂの重み係数Ｇ_ｂは、次式（２）に従って算出される。
【００５１】
【数２】

【００５２】
ここで、上式（２）中、ｓ_ｂ［ｎ］は、サブバンドｂの１次元合成フィルタ係数を示している。また、記号｜｜ｘ｜｜は、ベクトルｘに関するノルムを示す。
【００５３】
前記参考文献Ａに記載される数式（４．３９）と（４．４０）によれば、分解レベル１における低域成分Ｌ１の１次元合成フィルタ係数ｓ_Ｌ［１］［ｎ］と、同分解レベルにおける高域成分Ｈ１の１次元合成フィルタ係数ｓ_Ｈ［１］［ｎ］とは、次式（３）に従って算出される。
【００５４】
【数３】

【００５５】
ここで、上式（３）中、ｇ_０［ｎ］は、画像信号を帯域分割する順変換フィルタのローパス・フィルタ係数、ｇ_１［ｎ］は、そのハイパス・フィルタ係数をそれぞれ示している。
【００５６】
また、分解レベルｄ（ｄ＝１，２，…，Ｄ）における低域成分Ｌｄの１次元合成フィルタ係数ｓ_Ｌ［ｄ］［ｎ］と、同分解レベルにおける高域成分Ｈｄの１次元合成フィルタ係数ｓ_Ｈ［ｄ］［ｎ］とは、次式（４）に従って算出される。
【００５７】
【数４】

【００５８】
そして、分解レベルｄにおける低域成分Ｌｄの１次元合成フィルタ係数のノルムの二乗は、次式（５）に従って算出される。
【００５９】
【数５】

【００６０】
高域成分の１次元合成フィルタ係数のノルムの二乗も、上式（５）と同様にして算出することができる。
【００６１】
次に、分解レベルｄ（ｄ＝１，２，…，Ｄ；Ｄは整数）における帯域成分ＬＬＤ，ＨＬｄ，ＬＨｄ，ＨＨｄの２次元合成フィルタ係数は、上記１次元合成フィルタ係数の積で表現することができ、帯域成分ｂの２次元の重み係数Ｇ_ｂも、１次元の重み係数の積で表現することができる。具体的には、２次元合成フィルタ係数と２次元の重み係数とは、次式（６）に従って算出される。
【００６２】
【数６】

【００６３】
上式（６）中、添字ＬＬ［Ｄ］はサブバンドＬＬＤを示し，ＨＬ［ｄ］，ＬＨ［ｄ］およびＨＨ［ｄ］はそれぞれサブバンドＨＬｄ，ＬＨｄおよびＨＨｄを表している。
【００６４】
重み係数Ｇ_ｂの平方根がノルムである。以下の表１〜表４に、２次元の重み係数Ｇ_ｂに関する計算結果を示す。表１に、（９，７）フィルタ（９×７タップのフィルタ）の各帯域成分のノルムの二乗の数値を、表２には、表１に対応するノルムの数値をそれぞれ示す。また、表３に、（５，３）フィルタ（５×３タップのフィルタ）の各帯域成分のノルムの二乗の数値を、表４には、表３に対応するノルムの数値をそれぞれ示す。
【００６５】
【表１】

【００６６】
【表２】

【００６７】
【表３】

【００６８】
【表４】

【００６９】
更に、分解レベル１における低域成分ＬＬ１のノルムをαで表すとき、このノルムαを用いて、各帯域成分について図６に示すような値を設定する。図６に示す２次元画像２７は、オクターブ分割方式に従って帯域分割された２次元画像１２０を示す図である。分解レベルｎ（ｎ：１以上の整数）における帯域成分ＨＨｎの設定値は「２^ｎ−３×α」、帯域成分ＨＬｎおよびＬＨｎの設定値は「２^ｎ−２×α」、帯域成分ＬＬｎの設定値は「２^ｎ−１×α」にそれぞれ設定される。従って、例えば、帯域成分ＬＨ１の設定値は「２^−１×α」に設定されている。
【００７０】
上記設定値と、表２と表４に示したノルムの数値とを比較すれば、両者は概ね近似する。例えば、表２の場合（α＝１．９６５９１）、図６に示す各帯域成分の「設定値（対応する帯域成分）」は、約０．４９（ＨＨ１）、約０．９８（ＨＬ１，ＬＨ１）、約１．９６（ＨＬ２，ＬＨ２，ＨＨ３）、約３．９３（ＨＬ３，ＬＨ３）、約７．８６（ＬＬ３）となり、これら設定値は、表２に示したノルムの数値と近似していることが分かる。
【００７１】
また、図６において、帯域成分ＬＬ１のノルムをα＝２に丸め込み、各帯域成分の設定値を１ビット左シフトした値、すなわち、全ての設定値に２^１を乗算した２の巾乗値の指数は、図３に示した優先度の値に一致することが分かる。よって、第１実施例のように各帯域成分に優先度を設定することは、近似的に、レート・歪み最適化で使用するフィルタのノルム（重み係数の平方根）を各帯域成分のサンプル値（変換係数値）に乗算することに等しい。従って、本実施例の優先度は画像の歪みを低減させ得るように設定されるものである。
【００７２】
優先度設定処理（第２実施例）．
次に、優先度設定方法の第２実施例について説明する。本実施例では、各帯域成分の上記重み係数Ｇ_ｂの平方根であるノルムを、最も高い分解レベルにおける最低域成分ＬＬのノルムで除算した値を２の巾乗値に丸め込み、その２の巾乗値の指数の絶対値を、優先度として設定する。具体的には、最も高い分解レベルｎの最低域成分ＬＬｎのノルムをαとし、その他の帯域成分のノルムをｘとし、２の巾乗に丸め込む変数ｙに関する関数をＲ［ｙ］とし、変数ｙの２の巾乗２^ｍの指数ｍを算出する関数をｍ＝Ｉ［２^ｍ］とし、変数ｙに関する絶対値を｜ｙ｜とするとき、優先度ｐは、ｐ＝｜Ｉ［Ｒ［ｘ／α］］｜、に従って算出される。
【００７３】
以下の表５に、上記表２に示した（９，７）フィルタのノルムを用いて算出した優先度を示す。ここで、最も高い分解レベルは５であり、α＝３３．９２４９３、である。また、図７に、表５に示した優先度を記した２次元画像２８を示す帯域分割図を示す。尚、表中の「×」は、当該帯域成分の優先度は計算されていないことを意味する。
【００７４】
【表５】

【００７５】
また、以下の表６に、上記表４に示した（５，３）フィルタのノルムを用いて算出した優先度を示す。
【００７６】
【表６】

【００７７】
上記第１実施例では、各帯域成分の変換係数を優先度のビット数だけ左シフトさせて優先度を設定していたが、本第２実施例では、各帯域成分の変換係数を優先度のビット数だけ右シフトさせる処理が実行される。但し、変換係数のビット長を拡大させるように右ビットシフト処理が実行される。図８は、図７に示す優先度のビット数だけ右シフトされた帯域成分の変換係数２９，２９，…を示す模式図である。
【００７８】
優先度設定処理（第３実施例）．
次に、人間の視覚特性を考慮した第３実施例に係る優先度設定方法について説明する。数百万画素程度の高解像度の画像について上記第２実施例で示した優先度を適用した場合、復号画像の画質は客観評価では良いが、人間の視覚評価では必ずしも良いとは限らない。そこで、本実施例の優先度設定方法は、人間の視覚特性を考慮した重み付けをされた優先度を採用するものである。これにより、高い表示画質の圧縮画像を生成することが可能になる。
【００７９】
前記参考文献ＡのＣｈａｐｔｅｒ　１６には、ＣＳＦ（ｈｕｍａｎ　ｖｉｓｕａｌ　ｓｙｓｔｅｍ　Ｃｏｎｔｒａｓｔ　Ｓｅｎｓｉｔｉｖｉｔｙ　Ｆｕｎｃｔｉｏｎ）に基づいた重み付けＭＳＥ（Ｗｅｉｇｈｔｅｄ　Ｍｅａｎ　Ｓｑｕａｒｅｄ　Ｅｒｒｏｒ；ＷＭＳＥ）が記載されている。この記載によれば、人間の視覚評価を改善するために、上式（１）を次式（７）に修正するのが望ましい。
【００８０】
【数７】

【００８１】
ここで、上式（７）中、Ｗ_ｂ［ｉ］ ^ｃｓｆは、サブバンドｂ［ｉ］の”ｅｎｅｒｇｙ　ｗｅｉｇｈｔｉｎｇ　ｆａｃｔｏｒ”と呼ばれており、Ｗ_ｂ［ｉ］ ^ｃｓｆの推奨数値は、「ＩＳＯ／ＩＥＣ　ＪＴＣ　１／ＳＣ　２９／ＷＧ１（ＩＴＵ−Ｔ　ＳＧ８）　Ｎ２４０６，　”ＪＰＥＧ　２０００　Ｐａｒｔ　１　ＦＤＩＳ　（ｉｎｃｌｕｄｅｓ　ＣＯＲ　１，　ＣＯＲ　２，　ａｎｄ　ＤＣＯＲ３），”　４　Ｄｅｃｅｍｂｅｒ　２００１」の文献（以下、参考文献Ｂと呼ぶ。）に記載されている。図９〜図１１に、参考文献Ｂに記載される”ｅｎｅｒｇｙ　ｗｅｉｇｈｔｉｎｇ　ｆａｃｔｏｒ”の数値を示す。
【００８２】
図９〜図１１中の”ｌｅｖｅｌ”および　”Ｌｅｖ”は分解レベルを、”Ｃｏｍｐ”は輝度成分Ｙと色差成分Ｃｂ，Ｃｒをそれぞれ示しており、”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ（視距離）”が１０００，１７００，２０００，３０００，４０００の例が示されている。また、”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ　１０００”，　”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ　１７００”，　”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ　２０００”，　”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ　３０００”，　”Ｖｉｅｗｉｎｇ　ｄｉｓｔａｎｃｅ　４０００”は、それぞれ、１００ｄｐｉ，１７０ｄｐｉ，２００ｄｐｉ，３００ｄｐｉ，４００ｄｐｉのディスプレイまたは印刷物を１０インチ離れて見たときの視距離を意味する。
【００８３】
図９〜図１１に示す数値を用いて、上式（７）の重み付け係数の平方根（Ｗ_ｂ［ｉ］ ^ｃｓｆ・Ｇ_ｂ［ｉ］）^１／２を計算した。その計算結果を、以下の表７〜表１８に示す。表７〜表９は、図９に示す数値を用いて計算した（９，７）フィルタの白黒用数値、表１０〜表１２は、図１０と図１１に示す数値を用いて計算した（９，７）フィルタのカラー用数値を、表１３〜表１５は、図９に示す数値を用いて計算した（５，３）フィルタの白黒用数値を、表１６〜表１８は、図１０と図１１に示す数値を用いて計算した（５，３）フィルタのカラー用数値をそれぞれ示している。
【００８４】
【表７】

【００８５】
【表８】

【００８６】
【表９】

【００８７】
【表１０】

【００８８】
【表１１】

【００８９】
【表１２】

【００９０】
【表１３】

【００９１】
【表１４】

【００９２】
【表１５】

【００９３】
【表１６】

【００９４】
【表１７】

【００９５】
【表１８】

【００９６】
次に、上記表７〜表１８に示す数値を用いて、上記第２実施例で述べたのと同じ手順で各帯域成分の優先度を算出した。すなわち、最も高い分解レベルｎの最低域成分ＬＬｎの数値をαとし、その他の帯域成分の数値をｘとし、２の巾乗に丸め込む変数ｙに関する関数をＲ［ｙ］とし、変数ｙの２の巾乗２^ｍの指数ｍを算出する関数をｍ＝Ｉ［２^ｍ］とし、変数ｙに関する絶対値を｜ｙ｜とするとき、優先度ｐは、ｐ＝｜Ｉ［Ｒ［ｘ／α］］｜、に従って算出される。
【００９７】
優先度の値を、以下の表１９〜表３０に示す。表１９，表２０，表２１，表２２，表２３，表２４，表２５，表２６，表２７，表２８，表２９および表３０の優先度は、それぞれ、上記した表７，表８，表９，表１０，表１１，表１２，表１３，表１４，表１５，表１６，表１７および表１８の数値を用いて算出されたものである。
【００９８】
【表１９】

【００９９】
【表２０】

【０１００】
【表２１】

【０１０１】
【表２２】

【０１０２】
【表２３】

【０１０３】
【表２４】

【０１０４】
【表２５】

【０１０５】
【表２６】

【０１０６】
【表２７】

【０１０７】
【表２８】

【０１０８】
【表２９】

【０１０９】
【表３０】

【０１１０】
本実施例では、上記第２実施例と同じように、以上の表１９〜表３０に示す優先度のビット数だけ右シフトさせることで、各帯域成分の変換係数に対して優先度が設定される。これにより、人間の視覚特性を考慮した優先度を設定できる。
【０１１１】
画質制御処理．
次に、図２に示した画質制御部１０の構成と処理内容について説明する。図１２は、この画質制御部１０の概略構成を示す機能ブロック図である。
【０１１２】
この画質制御部１０は、目標画質（高画質，標準画質，低画質など）に基づいて、複数の画質パラメータ群から当該目標画質に適した画質パラメータＱＰを選択して出力する画質パラメータ選択部３１と、符号化対象を決定する符号化対象判定部３０とを備えている。符号化対象判定部３０は、優先度テーブル６から取得した優先度データＰＳ１に従って、入力ビットストリームに含まれる圧縮画像データの各帯域成分に対して上述の優先度を設定する。また符号化対象判定部３０は、設定した優先度に従って、前記画質パラメータＱＰで指定される目標画質に合わせて符号化対象を決定し、走査領域情報ＳＡを生成出力する。
【０１１３】
以下、符号化対象判定部３０における符号化対象の決定方法について説明する。図１３は、優先度に応じてビットシフトされた変換係数３３，３３，…を例示する模式図である。各変換係数３３は優先度に応じてビットシフトされている。また、変換係数３３の各ビットに付した番号０，１，…，１０は、当該ビットが属するビットプレーンの番号を示している。ここで、ＬＳＢ番号＝０，ＭＳＢ番号＝１０、である。
【０１１４】
符号化対象判定部３０は、画質パラメータＱＰに従って符号化終了ライン３２を設定し、当該符号化終了ライン３２よりも上位ビットを符号化対象に決定し、そのライン３２よりも下位ビットを符号化対象から外すように走査領域情報ＳＡを生成する。これにより符号化対象を効率的に選別することが可能になる。この結果、走査領域情報ＳＡを受けた符号量制御部１１は、各コードブロックにおいて、符号化終了ライン３２よりも上位のビットプレーンのみを走査し、そのライン３２よりも下位のビットプレーンを切り捨てることになる。
【０１１５】
符号化対象判定部３０は、更に、画質パラメータＱＰに従って、符号化パス単位で符号化対象を決定することができる。画質パラメータＱＰは、符号化対象のビットプレーンの制限と、符号化対象の符号化パス（ＣＬパス，ＳＩＧパスおよびＭＲパス）の制限とを示すパラメータ群を含んでいる。以下の表３１に、２０４８×２５６０画素の解像度をもつ画像に適した画質パラメータＱＰを例示する。尚、最低域のサブバンドの解像度を１２８×１２８画素よりも小さくする必要があるため、５以上の分解レベルが必要である。
【０１１６】
【表３１】

【０１１７】
表３１において「ビットプレーン数」は、図１３に示した符号化終了ライン３２よりも下位ビットの切り捨て対象のビットプレーンの数を、「パス名」は、符号化対象の中の最終符号化パスを、「最大パス数」は、符号化対象の符号化パス数の上限をそれぞれ表している。
【０１１８】
図１３と表３１を適用した場合の処理例を以下に説明する。図１４に、帯域成分ＬＬ５の変換係数３３として”０００１１０１０１１１_２＝２１５_１０”を例示する（Ｙ_２は２進値Ｙを、Ｘ_１０は１０進値Ｘを表すものとする）。表３１に示す通り、帯域成分ＬＬ５における最終符号化パスはＣＬパス、最大パス数は１７に制限されている。
【０１１９】
図１４に示す変換係数の７番目ビットは、ＳＩＧパスまたはＣＬパスに属するようにコンテクスト判定がなされている。８番目〜１０番目の上位ビットは、０ビットのみで構成されるビットプレーンに属する場合はタグツリー（Ｔａｇ　ｔｒｅｅ）と称する方式で符号化されており、既に符号化パスが開始している場合はＳＩＧパスまたはＣＬパスで符号化されている。７番目ビットが符号化開始パス（ＣＬパス）に属する場合、６番目ビットを含む下位ビットは、ＭＲパスに属するようにコンテクスト判定される。一般に、符号化開始のビットプレーンよりも下位のビットプレーンは、符号化効率の観点から、ＳＩＧパス，ＭＲパスおよびＣＬパスの順番で符号化されている。よって、最大パス数は１７に制限されているため、７番目ビットのＣＬパスから１番目ビットのＳＩＧパス迄の計１７パスが符号化対象になる。但し、１番目ビットはＭＲパスに属するため符号化対象に入らない。従って、下位２ビットは切り捨てられ、切り捨て後の値は”０００１１０１０１００_２＝２１２_１０”となる。この値がミッドポイントで逆量子化されれば、”０００１１０１０１１０_２＝２１４_１０”となる。
【０１２０】
次に、図１５に、帯域成分ＬＬ５の変換係数３３として”０００００００１１１１_２＝１５_１０”を例示する。変換係数の３番目ビットは、ＳＩＧパスまたはＣＬパスに属する。４番目〜１０番目の上位ビットは、０ビットのみで構成されるビットプレーンに属する場合はタグツリー（Ｔａｇ　ｔｒｅｅ）で符号化されており、既に符号化パスが開始している場合はＳＩＧパスまたはＣＬパスで符号化されている。３番目ビットが符号化開始パス（ＣＬパス）に属する場合、２番目ビットを含む下位ビットはＭＲパスに属し、３番目ビットのＣＬパスから０番目ビットのＣＬパス迄の計１０パスが符号化対象になる。切り捨て後の値は”０００００００１１１１_２＝１５_１０”となり、この値が逆量子化されれば、”０００００００１１１１_２＝１５_１０”となる。
【０１２１】
次に、図１６に、帯域成分ＨＨ２の変換係数３３として”００００１０１１１１１_２＝９５_１０”を例示する。表３１に示す通り、帯域成分ＨＨ２における最終符号化パスはＳＩＧパス、最大パス数は１４に制限されている。また下位３ビットのビットプレーンは切り捨てられる。
【０１２２】
変換係数の６番目ビットは、ＳＩＧパスまたはＣＬパスに属する。７番目〜１０番目の上位ビットは、０ビットのみで構成されるビットプレーンに属する場合はタグツリー（Ｔａｇ　ｔｒｅｅ）で符号化されており、既に符号化パスが開始している場合はＳＩＧパスまたはＣＬパスで符号化されている。６番目ビットが符号化開始パス（ＣＬパス）に属する場合、５番目ビットを含む下位ビットはＭＲパスに属する。また、３番目ビットプレーンのＳＩＧパス迄しか符号化しないという制限のため、６番目ビットのＣＬパスから４番目ビットのＳＩＧパス迄の８パスが符号化対象になるが、３番目ビットはＭＲパスに属するため符号化対象に入らない。従って、切り捨て後の値は”００００１０１００００_２＝８０_１０”となり、この値がミッドポイントで逆量子化されれば、”００００１０１１０００_２＝８８_１０”となる。
【０１２３】
尚、各ビットプレーンが、ＳＩＧパス，ＭＲパスおよびＣＬパスの順番で符号化されているのは、ＳＩＧパスの歪みに対する符号化効率が最も高いからである。図１７に、各符号化パスにおけるレート・歪み特性を示す。Ｒ−Ｄ曲線中、点Ｐ_１〜Ｐ_２の部分がＳＩＧパス，点Ｐ_２〜Ｐ_３の部分がＭＲパス、点Ｐ_３〜Ｐ_４の部分がＣＬパスを示している。各符号化パスにおけるレート（符号量）に対する歪みの比率ΔＤ_ＳＩＧ／ΔＲ_ＳＩＧ，ΔＤ_ＭＲ／ΔＲ_ＭＲ，ΔＤ_ＣＬ／ΔＲ_ＣＬをみれば、ＳＩＧパスにおける曲線勾配が最も急であり、符号化効率が最も高いことが分かる。
【０１２４】
以上のように、本実施形態に係る画質制御処理では、優先度に応じてビットシフトした変換係数に対して、変換係数を符号化対象とするか否かが決定される。符号化対象のみが選択されるため、歪みの少ない高画質の圧縮画像を生成し得るように、符号量を効率良く制御することが可能である。
【０１２５】
符号量制御処理．
次に、図２に示した符号量制御部１１の処理内容について説明する。符号量制御部１１は、入力ビットストリームに含まれる圧縮符号化データの容量の小計を、帯域成分単位，ビットプレーン単位および符号化パス単位で算出する。
【０１２６】
前記符号量制御部１１は、以下に説明する走査順序で並べ替えて生成した符号列から、目標符号量に適合するように切り捨て点（ｔｒｕｎｃａｔｉｏｎ　ｐｏｉｎｔ）を算出する。次に、符号量制御部１１は、その符号列のうち切り捨て点よりも前の符号列を読出すように読出制御信号ＣＳ１をＭＭＵ３に出力する。
【０１２７】
図１８と図１９は、前記走査順序と切り捨て点の一例を説明するための図である。図１８と図１９には、図１３に示したのと同じ規則で、優先度に応じてビットシフトされた変換係数３３，３３，…が表示されている。
【０１２８】
図１８の矢印に示すように、変換係数３３，３３，…は、ビットプレーン単位または符号化パス単位で、優先度の高い順に（上位ビットから下位ビットに向けて）且つ同一の優先度においては高域側から低域側に向けた走査順序で並べ替えられる。一般に、下位のビットプレーンを符号化する程にＭＲパスの割合が増えて圧縮効率が下がる傾向にある。よって、圧縮効率を向上させるために出来るだけ多くのＳＩＧパスを符号化すべく、同一の優先度においては高域側から低域側に向けた走査順序を採用している。
【０１２９】
そして、符号量制御部１１は、実際の符号量（バイト数）が目標符号量（バイト数）以下になる条件を満たすように切り捨て点を決定し、当該切り捨て点以降の符号列に含まれる下位ビットプレーンを切り捨てる。これにより、圧縮符号化データの符号量制御を、各サブバンドに設定した優先度に従って効率的に行うことができる。ここで、図１９に示すように、目標符号量に合わせてサブバンドＨＬ３の２番目ビットプレーンが切り捨て点として決定された場合、矢印で示す部分のビットが切り捨てられることになる。
【０１３０】
図２０は、ビットプレーン単位で並べ替えられた符号列を示す図、図２１は、符号化パス単位で並べ替えられた符号列を示す図である。図２０では、各ビットプレーンに対して、サブバンドを示す符号ＬＬ５，ＨＬ５，…と、ビットプレーン番号１０，９，…とが付されている。サブバンドＨＬ３の２番目ビットプレーンに付されたライン４４以降のビットプレーンが切り捨てられる。
【０１３１】
また、図２１では、各符号化パスに対して、符号化パスの種類を示す符号ＣＬ，ＳＩＧ，ＭＲと、サブバンドを示す符号ＬＬ５，ＨＬ５，…と、ビットプレーン番号１０，９，…とが付されている。サブバンドＨＬ３の２番目ビットプレーンのＭＲパスに付されたライン４４以降のビットプレーンが切り捨てられる。
【０１３２】
以上のように本実施形態に係る符号量制御処理によれば、レート・歪み最適化処理のために各符号化パスにおける歪量を利用せずに済み、リアルタイム性が高く、オーバーヘッドが低い高効率の符号量制御を実現できる。
【０１３３】
レイヤー分割処理．
次に、図２に示したレイヤー分割制御部７の動作を以下に説明する。レイヤー分割制御部７は、優先度テーブル６から取得した優先度データＰＳ２を用いて、入力ビットストリームに含まれる圧縮符号化データを優先度に対応するビット数だけビットシフトした符号列に変換し、その符号列を複数のレイヤー（マルチ・レイヤー）に分割させる制御機能を有している。
【０１３４】
以下、レイヤー分割処理について説明する。ＭＭＵ３は、入力ビットストリームを大容量の記憶装置２に一時的に記憶させる。レイヤー分割制御部７は、ＭＭＵ４１から圧縮符号化データのデータ構造情報ＤＳを取得する。次いで、レイヤー分割制御部７は、優先度テーブル６から優先度データＰＳ２を取得し、この優先度データＰＳ２に含まれる優先度に従って、圧縮符号化データの各帯域成分の変換係数を所定ビット数だけシフトさせる。これにより、各帯域成分の変換係数に対して優先度が設定される。優先度の設定方法としては、上記した第１実施例、第２実施例および第３実施例の方法を採用すればよい。
【０１３５】
図２２は、優先度に対応するビット数だけシフトされた変換係数４４，４４，…を例示する模式図である。各帯域成分ＬＬ５〜ＨＨ１の変換係数４４，４４，…は、優先度のビット数だけ右ビットシフト或いは左ビットシフトされている。また、各変換係数４４の各ビットに付した番号０，１，…，１０は、当該ビットが属するビットプレーンの番号を示している。ここで、ＬＳＢ番号＝０，ＭＳＢ番号＝１０、である。
【０１３６】
次に、レイヤー分割制御部７は、レイヤー分割情報に基づき、ビットシフトした符号化データを、ビットプレーン単位或いは符号化パス単位で複数のレイヤーにグループ分けするように分割位置を決定する。レイヤー分割情報としては、シングル・レイヤーとマルチ・レイヤーとの何れかを選択させる選択情報や、ビットプレーン単位または符号化パス単位でレイヤー分割位置を指定する情報などが含まれる。図２２の例では、圧縮符号化データを５枚のレイヤー０〜レイヤー４にビットプレーン単位で分割する分割位置が示されている。そして、レイヤー分割制御部７は、その分割位置に従ってレイヤー単位でデータを読出す旨の読出制御信号ＣＳ２をＭＭＵ３に供給する。ＭＭＵ３は、読出制御信号ＣＳ２に従って、記憶装置２に記憶されたデータＯＤを、上位レイヤーから下位レイヤーにかけて順番に読出して、多重化部５に出力する。
【０１３７】
以上のレイヤー分割処理では、優先度は、各帯域成分を当該優先度に対応するビット数だけビットシフトして設定される。このようにビットシフトした帯域成分を複数のレイヤーに分割することで、レートに対する歪みを低減し得るように、ビットプレーン単位或いは符号化パス単位で複数のレイヤーを効率的に生成することが可能である。従って、必ずしも、上述のレート・歪み最適化を用いてレイヤー分割処理を行う必要は無く、歪みを低減し得るようにリアルタイム性の高いレイヤー分割処理を行うことが可能になる。
【０１３８】
【発明の効果】
以上の如く、本発明の請求項１に係る符号量制御装置および請求項９に係るプログラムによれば、各帯域成分は優先度に対応するビット数だけシフトされ、シフトするビット数に応じて帯域成分の優先度が定まることから、圧縮画像の目標画質に合わせて符号化対象を効率的に指定し、少ない演算量で高速な符号量制御を行うことが可能になる。
【０１３９】
請求項２および請求項１０によれば、人間の視覚評価に適した、高い表示画質を実現するように符号量制御を行うことが可能となる。
【０１４０】
請求項３および請求項１１によれば、目標画質に合わせてビットプレーン単位で符号量を細かく且つ効率的に制御できる。
【０１４１】
請求項４および請求項１２によれば、目標画質に合わせて符号化パス単位で符号量を細かく且つ効率的に制御できる。
【０１４２】
請求項５，６および請求項１３，１４によれば、圧縮符号化データの符号量制御を、当該圧縮符号化データを復号化せずに、各帯域成分に設定した優先度に従って効率的に行うことが可能である。また、必ずしも、レート・歪み最適化を用いなくても、歪みを抑制し得るようにリアルタイム性の高い符号量制御を行うことが可能である。
【０１４３】
請求項７および請求項１５によれば、優先度に応じてビットシフトした帯域成分を複数のレイヤーに分割するため、レートに対する歪みを低減し得るように複数のレイヤーを効率的に生成することが可能である。
【０１４４】
請求項８および請求項１６によれば、人間の視覚評価に適した、高い表示画質を有する圧縮画像を生成することが可能となる。
【図面の簡単な説明】
【図１】本発明の実施形態に係る符号量制御装置の概略構成を示す機能ブロック図である。
【図２】図１に示した符号量制御装置におけるビット切り捨て制御部の概略構成を示す機能ブロック図である。
【図３】オクターブ分割方式に従って帯域分割した２次元画像を示す模式図である。
【図４】ビットシフトによる優先度設定処理を説明するための図である。
【図５】ビットシフトされた変換係数を例示する図である。
【図６】ウェーブレット変換によって帯域分割した２次元画像を示す模式図である。
【図７】ウェーブレット変換によって帯域分割した２次元画像を示す模式図である。
【図８】図７に示す優先度に応じて右ビットシフトされた帯域成分の変換係数を示す模式図である。
【図９】Ｅｎｅｒｇｙ　ｗｅｉｇｈｔｉｎｇ　ｆａｃｔｏｒの数値テーブルを示す図である。
【図１０】Ｅｎｅｒｇｙ　ｗｅｉｇｈｔｉｎｇ　ｆａｃｔｏｒの数値テーブルを示す図である。
【図１１】Ｅｎｅｒｇｙ　ｗｅｉｇｈｔｉｎｇ　ｆａｃｔｏｒの数値テーブルを示す図である。
【図１２】本実施形態に係る画質制御部の概略構成を示す機能ブロック図である。
【図１３】優先度に応じてビットシフトされた変換係数を例示する模式図である。
【図１４】帯域成分ＬＬ５の変換係数の処理例を説明するための図である。
【図１５】帯域成分ＬＬ５の変換係数の処理例を説明するための図である。
【図１６】帯域成分ＨＨ２の変換係数の符号化処理例を説明するための図である。
【図１７】レート・歪み特性の曲線を示す図である。
【図１８】走査順序の一例を説明するための図である。
【図１９】切り捨て点の一例を説明するための図である。
【図２０】ビットプレーン単位で並べ替えられた符号列を示す図である。
【図２１】符号化パス単位で並べ替えられた符号列を示す図である。
【図２２】複数のレイヤーに分割された符号化データを例示する模式図である。
【図２３】ＪＰＥＧ２０００方式による圧縮符号化装置の概略構成を示す機能ブロック図である。
【図２４】オクターブ分割方式に従って帯域分割された２次元画像を示す模式図である。
【図２５】複数のコードブロックに分解された２次元画像を示す模式図である。
【図２６】コードブロックを構成する複数枚のビットプレーンを示す模式図である。
【図２７】３種類の符号化パスを示す模式図である。
【符号の説明】
１　符号量制御装置
２　記憶装置
３　ＭＭＵ
４　ビット切り捨て制御部
５　多重化部
６　優先度テーブル
７　レイヤー分割制御部
１０　画質制御部
１１　符号量制御部[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a code amount control device that controls a code amount of a bit stream including a compressed image signal.
[0002]
[Prior art]
JPEG2000 (Joint Photographic Experts Group 2000) has been formulated by ISO (International Organization for Standardization) and ITU-T (International Telecommunication Union Telecommunication Standardization Sector) as a next-generation high-efficiency encoding scheme for image data. The JPEG2000 system has a superior function to the currently mainstream JPEG (Joint Photographic Experts Group) system, employs DWT (Discrete Wavelet Transform; Discrete Wavelet Transform) as an orthogonal transform, and employs bit entropy coding. It is characterized in that a method called EBCOT (Embedded \ Block \ Coding \ with \ Optimized \ Truncation) involving plane coding is employed.
[0003]
Hereinafter, the JPEG2000 compression encoding procedure will be outlined with reference to the compression encoding device (encoder) 100 shown in FIG.
[0004]
The image signal input to the compression encoding device 100 is subjected to DC level conversion by the DC level shift unit 102 as necessary, and then output to the color space conversion unit 103. Next, the color space conversion unit 103 converts the color space of the signal input from the DC level shift unit 102. Next, the tiling unit 104 divides the image signal input from the color space conversion unit 103 into a plurality of rectangular region components called “tiles” and outputs them to the DWT unit 105. The DWT unit 105 performs an integer or real DWT on the image signal input from the tiling unit 104 in tile units, and outputs a transform coefficient obtained as a result. In the DWT, a one-dimensional filter that divides a two-dimensional image signal into a high-frequency component (high-frequency component) and a low-frequency component (low-frequency component) is applied in the vertical direction and the horizontal direction. In the basic method of JPEG2000, an octave division method is adopted in which only the band components divided into the low frequency side in both the vertical direction and the horizontal direction are recursively divided. The number of recursively divided bands is called a decomposition level (decomposition @ level).
[0005]
FIG. 24 is a schematic diagram showing a two-dimensional image 120 on which DWT of decomposition level 3 has been performed in accordance with the octave division method. At the decomposition level 1, the two-dimensional image 120 is divided into four band components HH1, HL1, LH1, and LL1 (not shown) by sequentially applying the above-described one-dimensional filter in the vertical direction and the horizontal direction. You. Here, “H” indicates a high frequency component, and “L” indicates a low frequency component. For example, HL1 is a band component including a high-frequency component H in the horizontal direction and a low-frequency component L in the vertical direction at the decomposition level 1. By generalizing the notation, "XYn" (X and Y are H or L; n is an integer of 1 or more) represents a horizontal band component X and a vertical band component Y at a decomposition level n. The band component consists of
[0006]
Next, at the decomposition level 2, the low-frequency component LL1 is band-divided into HH2, HL2, LH2, and LL2 (not shown). Further, at the decomposition level 3, the low frequency component LL2 is band-divided into HH3, HL3, LH3 and LL3. FIG. 24 shows the arrangement of the band components HH1 to LL3 generated as described above.
[0007]
Next, the quantization unit 106 has a function of scalar-quantizing the transform coefficients output from the DWT unit 105 as necessary. The quantization unit 106 also has a function of performing a bit shift process by the ROI unit 107 to give priority to the image quality of a specified area (ROI; Region Of Interest). When performing the reversible (lossless) transformation, the scalar quantization in the quantization unit 106 is not performed. In the JPEG2000 system, two types of quantization means, scalar quantization by the quantization unit 106 and post-quantization (truncation) described later, are provided.
[0008]
Next, the transform coefficients output from the quantization unit 106 are sequentially subjected to block-based entropy encoding by the coefficient bit modeling unit 108 and the arithmetic encoding unit 109 according to the above-described EBCOT, and the code amount control unit 110 Is used to control the rate. Specifically, the coefficient bit modeling unit 108 divides the band component of the input transform coefficient into regions called “code blocks” of about 32 × 32 or 64 × 64, and further divides each code block into Decompose into a plurality of bit planes composed of a two-dimensional array.
[0009]
FIG. 25 is a schematic diagram showing a two-dimensional image 120 decomposed into a plurality of code blocks 121, 121, 121,. FIG. 26 shows n bit planes 122 constituting the code block 121.₀~ 122_n-1It is a schematic diagram which shows (n: natural number). As shown in FIG. 26, when the binary value 123 of the transform coefficient at one point in the code block 121 is “011... 0”, the bits forming the binary value 123 are bit planes 122, respectively._n-1, 122_n-2, 122_n-3, ..., 122₀Is decomposed to belong to Bit plane 122 in the figure_n-1Represents the most significant bit plane consisting of only the most significant bit (MSB) of the transform coefficient, and the bit plane 122₀Represents the least significant bit plane consisting of only the least significant bit (LSB).
[0010]
In addition, the coefficient bit modeling unit 108_kA context (context) determination of each bit in (k = 0 to n-1) is performed, and as shown in FIG. 27, a bit plane 122 is determined according to the significance (determination result) of each bit._kIs divided into three types of coding passes, that is, a CL pass (CLeanup @ pass), an MR pass (Magnitude @ Refinement @ pass), and a SIG pass (SIGnification @ propagation @ pass). The algorithm of the context determination for each encoding pass is defined in the JPEG2000 standard. According to this, "significant" means a state in which the coefficient of interest is known to be non-zero in the encoding process so far, and "not significant" means that the coefficient value is zero. , Or a state that may be zero.
[0011]
The coefficient bit modeling unit 108 includes a SIG pass (a coding pass of a non-significant coefficient around which a significant coefficient is present), an MR pass (a coding pass of a significant coefficient), and a CL pass (a SIG pass and a remainder not corresponding to the MR pass). The bit plane coding is executed in three types of coding passes (the coding pass of the coefficient information). Bit plane coding is performed by scanning the bits of each bit plane in 4-bit units from the most significant bit plane to the least significant bit plane, and determining whether a significant coefficient exists. The number of bit planes composed of only insignificant coefficients (0 bits) is recorded in the packet header, and actual encoding starts from the bit plane in which the significant coefficient first appears. The bit plane at the start of encoding is encoded only by the CL pass, and bit planes lower than the bit plane are sequentially encoded by the above three types of encoding passes.
[0012]
Next, the arithmetic coding unit 109 performs arithmetic coding on the coefficient sequence from the coefficient bit modeling unit 108 on a coding pass basis based on the result of the context determination using the MQ coder. Note that there is a mode in which the arithmetic encoding unit 109 performs a bypass process in which a part of the coefficient sequence input from the coefficient bit modeling unit 108 is not arithmetically encoded.
[0013]
Next, the code amount control unit 110 controls the final code amount by performing post-quantization for discarding lower-order bit planes of the code string output from the arithmetic coding unit 109. Then, the bit stream generation unit 111 generates a bit stream in which the code sequence output from the code amount control unit 110 and the additional information (header information, layer configuration, scalability, quantization table, and the like) are multiplexed, and is generated as a compressed image. Output.
[0014]
[Problems to be solved by the invention]
In the above-described compression encoding apparatus 100, the final code amount and the like are controlled so as to optimize the amount of distortion with respect to the encoding rate by using rate / distortion optimization (RD optimization). The rate / distortion optimization is performed using the amount of distortion in each coding path calculated in the above-described arithmetic coding stage. The algorithm for rate / distortion optimization is described in "David S. Taubman and Michael W. Marcellin," "JPEG2000 IMAGE COMPRESSION FUNDAMENTALS, STANDARDs AND AND PRACTICE", and "KLUDERA reference documents"; Is done.
[0015]
On the other hand, in a digital device such as a digital camera, the image quality, image size, code amount, scalability (hierarchy) of compressed image data may be changed. In such a case, the code amount and scalability of the compressed image are changed using a transcoder. In the JPEG2000 system, the scalability and code amount can be changed without decoding the compressed image signal. However, in the conventional JPEG baseline system, once the compressed image signal is decoded, the scalability and code amount are changed. Since it is necessary to change and execute compression encoding, the amount of calculation is large and the real-time property is low.
[0016]
Therefore, by using the transcoder of the JPEG2000 system, the scalability and the code amount of the compressed coded data in the bit stream generated by the compression coding device 100 can be changed with a small amount of calculation. However, the JPEG2000 transcoder cannot know the amount of distortion calculated by the compression encoding device 100, and therefore cannot easily control the bit stream code amount using rate / distortion optimization. There is. Even if a transcoder that performs code amount control using rate / distortion optimization can be realized, the amount of calculation of the code amount control process becomes large, and its real-time performance is reduced.
[0017]
In view of the above problems, it is an object of the present invention to control the code amount of a bit stream including compressed encoded data with a small amount of computation and at high speed so as to suppress distortion with respect to a rate. It is to provide a code amount control device.
[0018]
[Means for Solving the Problems]
In order to solve the above-mentioned problem, the invention according to claim 1 recursively divides an image signal into a high-frequency component and a low-frequency component by wavelet transform, calculates transform coefficients of a plurality of band components, and performs the transform. A code amount control device for controlling a code amount of compressed image data compressed by entropy coding of a coefficient, wherein each of the band components of the compressed image data is recursively divided into the low frequency components. An image quality control unit that performs bit shifting of the band component by the number of bits corresponding to the priority set according to the number of times, and selects an encoding target from the bit shifted band component in accordance with a target image quality. It is characterized by:
[0019]
The invention according to claim 2 is the code amount control device according to claim 1, wherein the image quality control unit has a function of using the priority weighted in consideration of human visual characteristics.
[0020]
The invention according to claim 3 is the code amount control device according to claim 1 or 2, wherein the image quality control unit has a function of determining the encoding target in units of the bit plane.
[0021]
The invention according to claim 4 is the code amount control device according to any one of claims 1 to 3, wherein the image quality control unit has a function of determining the encoding target in units of the encoding pass. Have
[0022]
Next, according to a fifth aspect of the present invention, an image signal is recursively divided into high-frequency components and low-frequency components by wavelet transform to calculate transform coefficients of a plurality of band components, and the transform coefficients are entropy-coded. A code amount control device for controlling the code amount of the compressed image data compressed and compressed, wherein the band component of the compressed image data is recursively divided into the low-frequency components according to the number of times of band division. The bit component is bit-shifted by the number of bits corresponding to the set priority, and the code sequence generated by rearranging the bit-shifted coded data of the band component in a predetermined scanning order is converted to a target code amount. A code amount control unit configured to calculate a suitable truncation point and to output the code string before the truncation point.
[0023]
The invention according to claim 6 is the code amount control device according to claim 5, wherein the code amount control unit transmits the coded data in the order of higher priority and at the same priority with a higher frequency band. The code string is generated by rearranging in the scanning order from the side toward the low frequency side.
[0024]
The invention according to claim 7 is the code amount control device according to any one of claims 1 to 6, wherein each of the band components of the compressed image data is recursively converted into the low band component. A layer division control unit that performs bit shift of the band component by the number of bits corresponding to the priority set according to the number of times of band division, and controls to divide the band component with the bit shift into a plurality of layers. Is further provided.
[0025]
The invention according to claim 8 is the code amount control device according to claim 7, wherein the layer division control unit has a function of using the priority weighted in consideration of human visual characteristics. It is.
[0026]
Next, an invention according to claim 9 is to recursively divide an image signal into a high-frequency component and a low-frequency component by a wavelet transform, calculate transform coefficients of a plurality of band components, and calculate the transform coefficients by entropy coding. A program for controlling the code amount of the compressed image data that has been compressed in a compressed manner, and for each of the band components of the compressed image data, Bit shifting the band component by the number of bits corresponding to the set priority, and causing the microprocessor to function as an image quality control unit that selects an encoding target from the bit shifted band component according to a target image quality. It is characterized by.
[0027]
A tenth aspect of the present invention is the program according to the ninth aspect, wherein the image quality control unit causes the microprocessor to function so as to use the priority weighted in consideration of human visual characteristics. is there.
[0028]
The invention according to claim 11 is the program according to claim 9 or 10, wherein the image quality control unit causes the microprocessor to function so as to determine the encoding target in units of the bit plane.
[0029]
According to a twelfth aspect of the present invention, there is provided the program according to any one of the ninth to eleventh aspects, wherein the image quality control unit determines the encoding target in units of the encoding pass. Function.
[0030]
Next, an invention according to claim 13 is to recursively divide an image signal into a high-frequency component and a low-frequency component by a wavelet transform, calculate transform coefficients of a plurality of band components, and calculate the transform coefficients by entropy coding. A program for controlling the code amount of the compressed image data that has been compressed in a compressed manner, and for each of the band components of the compressed image data, in accordance with the number of times the band is recursively divided into the low frequency components. From the code string generated by bit-shifting the band component by the number of bits corresponding to the set priority and rearranging the bit-shifted coded data of the band component in a predetermined scanning order, from the code string generated, Calculating a truncation point to be performed, and causing the microprocessor to function as a code amount control unit that controls to output the code string before the truncation point. That.
[0031]
The invention according to claim 14 is the program according to claim 13, wherein the code amount control unit transmits the coded data in descending order of the priority and at the same priority from the high-frequency side to the low-frequency side. And causing the microprocessor to function so as to generate the code string by rearranging in the scanning order toward the area side.
[0032]
The invention according to claim 15 is the program according to any one of claims 9 to 14, wherein each of the band components of the compressed image data is recursively divided into the low-frequency components. The microcontroller is a layer division control unit that performs bit shifting of the band component by the number of bits corresponding to the priority set according to the number of times performed, and controls the band component to be divided into a plurality of layers. It makes the processor work.
[0033]
The invention according to claim 16 is the program according to claim 15, wherein the layer division control unit functions the microprocessor so as to use the weighted priority in consideration of human visual characteristics. It is to let.
[0034]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described.
[0035]
Configuration of code amount control device.
FIG. 1 is a functional block diagram illustrating a schematic configuration of a code amount control device according to an embodiment of the present invention. The code amount control device (transcoder) 1 includes a large-capacity storage device 2, an MMU (memory management unit) 3 that controls reading and writing of data from and to the storage device 2, a bit truncation control unit 4, And a priority table 6, and a layer division control unit 7.
[0036]
Note that all or a part of the processing units 4, 5, 6, and 7 constituting the code amount control device 1 may be constituted by hardware or may be constituted by a program for causing a microprocessor to function. .
[0037]
The MMU 3 of the code amount control device 1 temporarily stores, in the storage device 2, an input bit stream including compressed and encoded data compressed and encoded by the JPEG2000 method, and then controls the code amount according to control signals CS1 and CS2. The data OD is read from the storage device 2 and output to the multiplexing unit 5. The multiplexing unit 5 multiplexes the data OD and outputs it as an output bit stream.
[0038]
FIG. 2 is a functional block diagram showing a schematic configuration of the bit truncation control unit 4 in the code amount control device 1 shown in FIG. The bit truncation control unit 4 includes an image quality control unit 10 that selects an encoding target according to a target image quality, and a code amount control unit 11 that controls the code amount according to a target code amount (final code amount). . Further, the code amount control unit 11 calculates a truncation point suitable for the target code amount from the encoding target selected by the image quality control unit 10 based on the data structure information DS of the input bit stream supplied from the MMU 3. , Generates a read control signal CS1 for reading a code string before the truncation point, and supplies the read control signal CS1 to the MMU 3.
[0039]
The layer division control unit 7 shown in FIG. 1 generates a read control signal CS2 for generating an output bit stream divided into a plurality of layers based on the data structure information DS of the input bit stream supplied from the MMU 3. And outputs it to MMU3.
[0040]
The priority table 6 indicates the priority set for each band component of the compression-encoded data included in the input bit stream in accordance with the number of recursive band divisions into low-frequency components according to the JPEG2000 method. The priority data PS1 and PS2 are supplied to the bit truncation control unit 4 and the layer division control unit 7.
[0041]
The configuration and operation of the code amount control device 1 having the above configuration will be described in detail below.
[0042]
Priority setting processing (first embodiment).
A first embodiment of a method of setting the priority to be recorded in the priority table 6 will be described. In the present invention, the priority is determined according to the number of times that each band component (sub-band) is recursively divided into low-frequency components. In this embodiment, the priority of the band component HHn at the decomposition level n (n: an integer of 1 or more) is “n−1”, the priority of the band components HLn and LHn is “(n−1) +1”, and the band component is The priority of LLn is determined to be “(n−1) +2”. For example, the priority of the band component HH1 shown in FIG. 24 is set to “0”, and the priority of the band component LL3 is set to “4”. FIG. 3 is a schematic diagram showing a two-dimensional image 25 obtained by band division according to the octave division method. Each of the band components is assigned one of the priorities “0”, “1”, “2”, “3”, and “4”.
[0043]
The priority table 6 records priority information corresponding to each of the band components HHn, HLn, LHn, and LLn. The image quality control unit 10, the code amount control unit 11, and the layer division control unit 7 According to the priority data PS1 and PS2 acquired from the priority table 6, the priority is set for each band component. Specifically, the transform coefficients (hereinafter simply referred to as “transform coefficients”) entropy-encoded in each band component are shifted by the number of bits corresponding to the priorities, thereby giving priority to each transform coefficient. The degree is set. In the bit shift processing, it is not always necessary to actually perform a bit shift operation on each transform coefficient, and it is sufficient to virtually shift the position of each bit of the transform coefficient. In this case, the position of the bit plane to which each bit of the transform coefficient belongs does not change.
[0044]
FIG. 4 is a diagram for explaining the priority setting process by the bit shift. In the example shown in FIG. 3, since the priority of the band component LL3 is “4”, the corresponding transform coefficient 26 is shifted left by 4 bits. The conversion coefficients 26, 26 of the band components HL3, LH3 set to the priority "3" are shifted left by 3 bits, and the conversion coefficients 26, 26 of the band components HH3, HL2, LH2 set to the priority "2". , 26, 26 are shifted left by 2 bits, and the transform coefficients 26, 26, 26 of the band components HH2, HL1, LH1 to which the priority "1" is set are shifted left by 1 bit. At this time, as shown in FIG. 5, the conversion coefficient of the two-dimensional image 25A before the bit shift is changed to the conversion coefficient indicated by the two-dimensional image 25B by the above-described left bit shift processing. For example, the transform coefficient value (= 4) of the band component LL3 is 4 × 2 by shifting left by 4 bits.⁴= 64.
[0045]
Next, the reason for setting the priority as described above (theoretical background) will be described below.
[0046]
In the above-described conventional rate / distortion optimization (RD-optimization) method, an optimization process using a distortion measure is performed. David @ S.に よ According to the above-mentioned reference A by Taubman et al., The strain measure D_i ^(Z)Is calculated according to the following equation (1).
[0047]
(Equation 1)

[0048]
In the above formula (1), z is a bit truncation point (bit @ truncation @ point); oy_i ^{K [i, j]}[J] is the j-th sample value (coefficient value) of the code block dequantized by the K [i, j] -th bit plane; y_i[J] is the j-th sample value (coefficient value) of the code block; G_{b [i]}Is the square of the norm of the synthesis filter coefficient corresponding to the sub-band b [i], and indicates the weight coefficient of the distortion model depending on the sub-band b. Note that, for convenience of description, the notation of the symbol shown in the above equation (1) is slightly different from that in Reference A described above.
[0049]
In the rate / distortion optimization, this distortion measure D_i ^(Z)An optimization process is performed to minimize the total amount in the subband b [i]. Subband b weighting factor G_bRepresents weighting for reducing image distortion.
[0050]
Subband b weighting factor G_bIs calculated according to the following equation (2).
[0051]
(Equation 2)

[0052]
Here, in the above equation (2), s_b[N] indicates a one-dimensional synthesis filter coefficient of the subband b. The symbol || x || indicates a norm for the vector x.
[0053]
According to the equations (4.39) and (4.40) described in Reference A, the one-dimensional synthesis filter coefficient s of the low-frequency component L1 at the decomposition level 1 is obtained._{L [1]}[N] and a one-dimensional synthesis filter coefficient s of the high-frequency component H1 at the same decomposition level_{H [1]}[N] is calculated according to the following equation (3).
[0054]
(Equation 3)

[0055]
Here, in the above equation (3), g₀[N] is a low-pass filter coefficient of a forward conversion filter that divides an image signal into bands, and g₁[N] indicates the respective high-pass filter coefficients.
[0056]
The one-dimensional synthesis filter coefficient s of the low-frequency component Ld at the decomposition level d (d = 1, 2,..., D)_{L [d]}[N] and a one-dimensional synthesis filter coefficient s of the high frequency component Hd at the same decomposition level_{H [d]}[N] is calculated according to the following equation (4).
[0057]
(Equation 4)

[0058]
Then, the square of the norm of the one-dimensional synthesis filter coefficient of the low-frequency component Ld at the decomposition level d is calculated according to the following equation (5).
[0059]
(Equation 5)

[0060]
The square of the norm of the one-dimensional synthesis filter coefficient of the high-frequency component can be calculated in the same manner as in the above equation (5).
[0061]
Next, the two-dimensional synthesis filter coefficients of the band components LLD, HLd, LHd, and HHd at the decomposition level d (d = 1, 2,..., D; D is an integer) are expressed by the product of the one-dimensional synthesis filter coefficients. And the two-dimensional weighting factor G of the band component b_bCan also be expressed by the product of the one-dimensional weight coefficients. Specifically, the two-dimensional synthesis filter coefficient and the two-dimensional weight coefficient are calculated according to the following equation (6).
[0062]
(Equation 6)

[0063]
In the above equation (6), the subscript LL [D] indicates the sub-band LLD, and HL [d], LH [d] and HH [d] indicate the sub-bands HLd, LHd and HHd, respectively.
[0064]
Weight coefficient G_bIs the norm. Tables 1 to 4 below show two-dimensional weighting factors G_bThe calculation results for Table 1 shows numerical values of the square of the norm of each band component of the (9, 7) filter (a filter of 9 × 7 taps), and Table 2 shows numerical values of the norm corresponding to Table 1. Further, Table 3 shows the numerical value of the square of the norm of each band component of the (5, 3) filter (filter of 5 × 3 taps), and Table 4 shows the numerical value of the norm corresponding to Table 3.
[0065]
[Table 1]

[0066]
[Table 2]

[0067]
[Table 3]

[0068]
[Table 4]

[0069]
Further, when the norm of the low-frequency component LL1 at the decomposition level 1 is represented by α, a value as shown in FIG. 6 is set for each band component using the norm α. A two-dimensional image 27 shown in FIG. 6 is a diagram showing a two-dimensional image 120 that is band-divided according to the octave division method. The set value of the band component HHn at the decomposition level n (n: an integer of 1 or more) is “2^n-3× α ”, and the set values of the band components HLn and LHn are“ 2^n-2× α ”and the set value of the band component LLn is“ 2^n-1× α ”. Therefore, for example, the set value of the band component LH1 is “2^-1× α ”.
[0070]
When the set values are compared with the norm values shown in Tables 2 and 4, they are approximately similar. For example, in the case of Table 2 (α = 1.96591), the “set value (corresponding band component)” of each band component shown in FIG. 6 is about 0.49 (HH1) and about 0.98 (HL1, LH1). ), About 1.96 (HL2, LH2, HH3), about 3.93 (HL3, LH3), about 7.86 (LL3). These set values are approximate to the norm values shown in Table 2. I understand that there is.
[0071]
In FIG. 6, the norm of the band component LL1 is rounded to α = 2, and the set value of each band component is shifted left by one bit, that is, 2 is added to all the set values.¹It can be seen that the exponent of the power of 2 obtained by multiplying by 2 matches the priority value shown in FIG. Therefore, setting the priority to each band component as in the first embodiment is approximately equivalent to setting the norm (square root of the weighting factor) of the filter used in the rate / distortion optimization to the sample value of each band component ( Transform coefficient value). Therefore, the priority of this embodiment is set so as to reduce image distortion.
[0072]
Priority setting processing (second embodiment).
Next, a second embodiment of the priority setting method will be described. In this embodiment, the weight coefficient G of each band component_bIs divided by the norm of the lowest band component LL at the highest decomposition level, the result is rounded to a power of two, and the absolute value of the exponent of the power of two is set as the priority. Specifically, the norm of the lowest band component LLn of the highest decomposition level n is α, the norm of the other band components is x, the function related to the variable y rounded to the power of 2 is R [y], and the variable y 2 power of 2^mThe function for calculating the index m of m is m = I [2^mAnd the absolute value for the variable y is | y |, the priority p is calculated according to p = | I [R [x / α]] |.
[0073]
Table 5 below shows the priorities calculated using the norms of the (9, 7) filters shown in Table 2 above. Here, the highest decomposition level is 5, and α = 33.92493. FIG. 7 is a band division diagram showing a two-dimensional image 28 in which the priorities shown in Table 5 are described. Note that “x” in the table means that the priority of the band component has not been calculated.
[0074]
[Table 5]

[0075]
Table 6 below shows the priority calculated using the norm of the (5, 3) filter shown in Table 4 above.
[0076]
[Table 6]

[0077]
In the first embodiment, the priority is set by shifting the conversion coefficient of each band component to the left by the number of bits of the priority. In the second embodiment, the conversion coefficient of each band component is set to the priority. A process of shifting right by the number of bits is executed. However, the right bit shift process is performed so as to increase the bit length of the transform coefficient. FIG. 8 is a schematic diagram showing the conversion coefficients 29, 29,... Of the band components right-shifted by the number of bits of the priority shown in FIG.
[0078]
Priority setting processing (third embodiment).
Next, a priority setting method according to the third embodiment in consideration of human visual characteristics will be described. When the priority shown in the second embodiment is applied to a high-resolution image of about several million pixels, the image quality of a decoded image is good for objective evaluation, but not always good for human visual evaluation. Therefore, the priority setting method of the present embodiment employs weighted priorities in consideration of human visual characteristics. This makes it possible to generate a compressed image with high display quality.
[0079]
Chapter 16 of Reference A describes a weighted MSE (weighted mean squared error; WMSE) based on the CSF (human visual system contrast contrast sensitivity function). According to this description, it is desirable to modify the above equation (1) to the following equation (7) in order to improve human visual evaluation.
[0080]
(Equation 7)

[0081]
Here, in the above equation (7), W_{b [i]} ^csfIs called "energy @ weighting @ factor" of subband b [i], and W_{b [i]} ^csfThe recommended numerical values of “ISO / IEC TCC1 / SC 29 / WG1 (ITU-T SG8) N2406,“ JPEG 2000 Part 1 FDIS (includes COR 1, COR 2, and DCOR3), ”4December 2001 (2001) Hereinafter, referred to as Reference Document B.). 9 to 11 show numerical values of "energy @ weighting @ factor" described in Reference Document B.
[0082]
9 to 11, “level” and “Lev” indicate the decomposition level, “Comp” indicates the luminance component Y and the color difference components Cb and Cr, respectively, and “Viewing distance” (viewing distance) is 1000 or 1700. , 2000, 3000, 4000 are shown. Also, “Viewing distance 1000”, “Viewing distance 1700”, “Viewing distance 2000”, “Viewing distance 3000”, “Viewing distance 4000”, 100 dpi, 170 dpi, 100 dpi, 170 dpi, 100 dpi, and 100 dpi, 100 dpi, 170 dpi, 170 dpi, and 100 dpi, respectively. The viewing distance when viewed from an inch.
[0083]
Using the numerical values shown in FIGS. 9 to 11, the square root (W_{b [i]} ^csf・ G_{b [i]})^1/2Was calculated. The calculation results are shown in Tables 7 to 18 below. Tables 7 to 9 are numerical values for black and white of the (9, 7) filter calculated using the numerical values shown in FIG. 9, and Tables 10 to 12 are calculated using the numerical values shown in FIG. 10 and FIG. , 7) The color values of the filter are calculated using the values shown in FIG. 9 in Tables 13 to 15, and the black and white values of the (5, 3) filter are shown in FIGS. 11 shows the numerical values for the color of the (5, 3) filter calculated using the numerical values shown in FIG.
[0084]
[Table 7]

[0085]
[Table 8]

[0086]
[Table 9]

[0087]
[Table 10]

[0088]
[Table 11]

[0089]
[Table 12]

[0090]
[Table 13]

[0091]
[Table 14]

[0092]
[Table 15]

[0093]
[Table 16]

[0094]
[Table 17]

[0095]
[Table 18]

[0096]
Next, using the numerical values shown in Tables 7 to 18, the priority of each band component was calculated in the same procedure as described in the second embodiment. That is, the numerical value of the lowest band component LLn of the highest decomposition level n is α, the numerical values of the other band components are x, the function related to the variable y rounded to the power of 2 is R [y], and the variable y of the variable y is 2 Width 2^mThe function for calculating the index m of m is m = I [2^mAnd the absolute value for the variable y is | y |, the priority p is calculated according to p = | I [R [x / α]] |.
[0097]
The priority values are shown in Tables 19 to 30 below. The priorities of Table 19, Table 20, Table 21, Table 22, Table 23, Table 24, Table 25, Table 26, Table 27, Table 28, Table 29, and Table 30 are the same as those in Tables 7, 8, and 30, respectively. It is calculated using the numerical values of Tables 9, 9, 10, 11, 12, 13, 14, 15, 16, 17, and 18.
[0098]
[Table 19]

[0099]
[Table 20]

[0100]
[Table 21]

[0101]
[Table 22]

[0102]
[Table 23]

[0103]
[Table 24]

[0104]
[Table 25]

[0105]
[Table 26]

[0106]
[Table 27]

[0107]
[Table 28]

[0108]
[Table 29]

[0109]
[Table 30]

[0110]
In this embodiment, as in the second embodiment, the priority is set for the transform coefficient of each band component by right-shifting by the number of bits of the priority shown in Tables 19 to 30 above. You. As a result, the priority can be set in consideration of the human visual characteristics.
[0111]
Image quality control processing.
Next, the configuration and processing contents of the image quality control unit 10 shown in FIG. 2 will be described. FIG. 12 is a functional block diagram showing a schematic configuration of the image quality control unit 10.
[0112]
The image quality control unit 10 selects an image quality parameter QP suitable for the target image quality from a plurality of image quality parameter groups based on the target image quality (high image quality, standard image quality, low image quality, etc.) and outputs the selected image quality parameter QP. And an encoding target determination unit 30 that determines an encoding target. The encoding target determination unit 30 sets the above-described priority for each band component of the compressed image data included in the input bit stream according to the priority data PS1 acquired from the priority table 6. Further, the encoding target determining unit 30 determines an encoding target in accordance with the target image quality specified by the image quality parameter QP according to the set priority, and generates and outputs scanning area information SA.
[0113]
Hereinafter, a method of determining a coding target in the coding target determination unit 30 will be described. FIG. 13 is a schematic diagram illustrating the conversion coefficients 33, 33,... Bit-shifted according to the priority. Each transform coefficient 33 is bit-shifted according to the priority. .., 10 given to each bit of the conversion coefficient 33 indicate the number of a bit plane to which the bit belongs. Here, the LSB number = 0 and the MSB number = 10.
[0114]
The encoding target determination unit 30 sets an encoding end line 32 according to the image quality parameter QP, determines a higher-order bit than the encoding end line 32 as an encoding target, and sets a lower-order bit than the line 32 as an encoding target. The scanning area information SA is generated so as to be out of the range. This makes it possible to efficiently select an encoding target. As a result, the code amount control unit 11 having received the scanning area information SA scans only the bit plane higher than the coding end line 32 in each code block, and discards the bit plane lower than the line 32. become.
[0115]
The encoding target determination unit 30 can further determine the encoding target for each encoding pass according to the image quality parameter QP. The image quality parameter QP includes a parameter group indicating a restriction on a bit plane to be coded and a restriction on a coding pass (CL pass, SIG pass, and MR pass) to be coded. Table 31 below exemplifies image quality parameters QP suitable for images having a resolution of 2048 × 2560 pixels. Since the resolution of the lowest sub-band needs to be smaller than 128 × 128 pixels, a resolution level of 5 or more is required.
[0116]
[Table 31]

[0117]
In Table 31, “the number of bit planes” indicates the number of bit planes whose lower bits are lower than the encoding end line 32 shown in FIG. 13, and “path name” indicates the last encoding path in the encoding target. And the “maximum number of passes” represents the upper limit of the number of encoding passes to be encoded.
[0118]
An example of processing when FIG. 13 and Table 31 are applied will be described below. FIG. 14 shows “00011010111” as the conversion coefficient 33 of the band component LL5.₂= 215₁₀"(Y₂Represents the binary value Y, X₁₀Represents a decimal value X). As shown in Table 31, the final encoding pass in the band component LL5 is limited to a CL pass, and the maximum number of passes is limited to 17.
[0119]
The context determination is made so that the seventh bit of the transform coefficient shown in FIG. 14 belongs to the SIG pass or the CL pass. If the eighth to tenth upper bits belong to a bit plane consisting of only 0 bits, they are coded by a method called tag tree (Tag @ tree). If the coding pass has already started, It is encoded by the SIG pass or the CL pass. If the seventh bit belongs to the encoding start pass (CL pass), the lower bits including the sixth bit are context-determined to belong to the MR pass. Generally, bit planes lower than the bit plane at the start of encoding are encoded in the order of SIG pass, MR pass, and CL pass from the viewpoint of encoding efficiency. Therefore, since the maximum number of passes is limited to 17, a total of 17 passes from the 7th bit CL pass to the 1st bit SIG pass are to be encoded. However, the first bit is not included in the encoding target because it belongs to the MR pass. Therefore, the lower 2 bits are truncated, and the value after the truncation is “00011010100”.₂= 212₁₀If this value is inversely quantized at the midpoint, “00011010110” is obtained.₂= 214₁₀".
[0120]
Next, FIG. 15 shows “00000001111” as the conversion coefficient 33 of the band component LL5.₂= 15₁₀The third bit of the transform coefficient belongs to the SIG pass or the CL pass. If the fourth to tenth upper bits belong to a bit plane composed of only 0 bits, the tag tree (Tag @ tree) If the coding pass has already started, the coding has been performed by the SIG pass or the CL pass.If the third bit belongs to the coding start pass (CL pass), the second bit is The lower bits included belong to the MR pass, and a total of 10 passes from the CL pass of the 3rd bit to the CL pass of the 0th bit are to be encoded, and the value after truncation is “00000001111”.₂= 15₁₀And if this value is inversely quantized, it becomes “00000001111”.₂= 15₁₀".
[0121]
Next, FIG. 16 shows “00001011111” as the conversion coefficient 33 of the band component HH2.₂= 95₁₀As shown in Table 31, the final encoding path in the band component HH2 is limited to the SIG path, and the maximum number of paths is limited to 14. The lower three bit planes are truncated.
[0122]
The sixth bit of the transform coefficient belongs to the SIG pass or the CL pass. The seventh to tenth upper bits are coded by a tag tree (Tag @ tree) if they belong to a bit plane composed of only 0 bits, and if the coding pass has already started, the SIG pass or the Encoded by CL pass. When the sixth bit belongs to the coding start pass (CL pass), the lower bits including the fifth bit belong to the MR pass. In addition, due to the restriction that only the SIG pass of the third bit plane is encoded, eight passes from the CL pass of the sixth bit to the SIG pass of the fourth bit are to be encoded, but the third bit is an MR pass. And does not enter the encoding target. Therefore, the value after truncation is “0000001010000”.₂= 80₁₀And if this value is inversely quantized at the midpoint, "00001011000"₂= 88₁₀".
[0123]
The reason why each bit plane is encoded in the order of the SIG pass, the MR pass, and the CL pass is that the encoding efficiency for the distortion of the SIG pass is the highest. FIG. 17 shows the rate / distortion characteristics in each coding path. In the RD curve, point P₁~ P₂Is the SIG pass, point P₂~ P₃Is the MR pass, point P₃~ P₄Indicates a CL path. Ratio of distortion to rate (code amount) ΔD in each encoding pass_SIG/ ΔR_SIG, ΔD_MR/ ΔR_MR, ΔD_CL/ ΔR_CLIt can be seen that the curve slope in the SIG path is the steepest and the coding efficiency is the highest.
[0124]
As described above, in the image quality control processing according to the present embodiment, it is determined whether or not a transform coefficient is to be coded for a transform coefficient that has been bit-shifted according to the priority. Since only the encoding target is selected, the code amount can be efficiently controlled so that a high-quality compressed image with little distortion can be generated.
[0125]
Code amount control processing.
Next, the processing content of the code amount control unit 11 shown in FIG. 2 will be described. The code amount control unit 11 calculates the subtotal of the capacity of the compressed and coded data included in the input bit stream in units of band components, bit planes, and coding paths.
[0126]
The code amount control unit 11 calculates a truncation point (truncation point) from the code string generated by rearranging in the scanning order described below so as to conform to the target code amount. Next, the code amount control unit 11 outputs a read control signal CS1 to the MMU 3 so as to read a code string before the cut-off point in the code string.
[0127]
FIG. 18 and FIG. 19 are diagrams for explaining an example of the scanning order and the truncation point. In FIG. 18 and FIG. 19, the conversion coefficients 33, 33,... Which are bit-shifted according to the priority according to the same rule as shown in FIG. 13 are displayed.
[0128]
As shown by the arrows in FIG. 18, the transform coefficients 33, 33,... Are in the order of higher priority (from the upper bit to the lower bit) in units of bit planes or in units of coding paths and at the same priority. Rearrangement is performed in the scanning order from the high frequency side to the low frequency side. Generally, as the lower bit plane is encoded, the ratio of the MR pass increases, and the compression efficiency tends to decrease. Therefore, in order to encode as many SIG passes as possible in order to improve the compression efficiency, the scanning order from the high frequency side to the low frequency side is adopted at the same priority.
[0129]
Then, the code amount control unit 11 determines the truncation point so that the actual code amount (the number of bytes) is equal to or less than the target code amount (the number of bytes), and determines the lower-order point included in the code string after the cut-off point. Truncate the bit plane. Thereby, the code amount control of the compression encoded data can be efficiently performed according to the priority set for each subband. Here, as shown in FIG. 19, when the second bit plane of the sub-band HL3 is determined as a truncation point in accordance with the target code amount, the bits indicated by arrows are truncated.
[0130]
FIG. 20 is a diagram illustrating a code string rearranged in a unit of a bit plane, and FIG. 21 is a diagram illustrating a code string rearranged in a unit of an encoding pass. 20, symbols LL5, HL5,... Indicating subbands and bit plane numbers 10, 9,. The bit planes after the line 44 attached to the second bit plane of the subband HL3 are discarded.
[0131]
In FIG. 21, codes CL, SIG, and MR indicating the types of the coding paths, codes LL5, HL5,... Indicating the subbands, and bit plane numbers 10, 9,. Is attached. The bit planes after the line 44 attached to the MR pass of the second bit plane of the subband HL3 are discarded.
[0132]
As described above, according to the code amount control processing according to the present embodiment, it is not necessary to use the amount of distortion in each encoding path for the rate / distortion optimization processing, and the real-time property is high, the overhead is low, and the efficiency is high. Code amount control can be realized.
[0133]
Layer division processing.
Next, the operation of the layer division control unit 7 shown in FIG. 2 will be described below. Using the priority data PS2 acquired from the priority table 6, the layer division control unit 7 converts the compression-encoded data included in the input bit stream into a code string that is bit-shifted by the number of bits corresponding to the priority. It has a control function of dividing the code string into a plurality of layers (multi-layers).
[0134]
Hereinafter, the layer division processing will be described. The MMU 3 temporarily stores the input bit stream in the large-capacity storage device 2. The layer division control unit 7 acquires the data structure information DS of the compression-encoded data from the MMU 41. Next, the layer division control unit 7 acquires the priority data PS2 from the priority table 6, and according to the priority included in the priority data PS2, converts the conversion coefficient of each band component of the compression-encoded data by a predetermined number of bits. Shift. As a result, the priority is set for the transform coefficient of each band component. As a method of setting the priority, the methods of the first, second, and third embodiments described above may be employed.
[0135]
FIG. 22 is a schematic diagram illustrating conversion coefficients 44, 44,... Shifted by the number of bits corresponding to the priority. The transform coefficients 44, 44,... Of each of the band components LL5 to HH1 have been shifted right or left by the number of bits of priority. .., 10 given to each bit of each conversion coefficient 44 indicate the number of a bit plane to which the bit belongs. Here, the LSB number = 0 and the MSB number = 10.
[0136]
Next, based on the layer division information, the layer division control unit 7 determines a division position so that the bit-shifted encoded data is grouped into a plurality of layers in units of bit planes or in units of encoding paths. The layer division information includes selection information for selecting one of a single layer and a multi-layer, information for specifying a layer division position in units of bit planes or in units of encoding paths, and the like. In the example of FIG. 22, division positions at which the compression-encoded data is divided into five layers 0 to 4 in bit plane units are shown. Then, the layer division control unit 7 supplies the MMU 3 with a read control signal CS2 for reading data in layer units according to the division position. The MMU 3 sequentially reads the data OD stored in the storage device 2 from the upper layer to the lower layer according to the read control signal CS2, and outputs the data OD to the multiplexing unit 5.
[0137]
In the above-described layer division processing, the priority is set by bit-shifting each band component by the number of bits corresponding to the priority. By dividing the bit component that has been bit-shifted into a plurality of layers as described above, it is possible to efficiently generate a plurality of layers in units of bit planes or in units of encoding paths so that distortion with respect to a rate can be reduced. is there. Therefore, it is not always necessary to perform the layer division processing using the above-described rate / distortion optimization, and it is possible to perform the layer division processing with high real-time property so as to reduce distortion.
[0138]
【The invention's effect】
As described above, according to the code amount control device according to the first aspect of the present invention and the program according to the ninth aspect, each band component is shifted by the number of bits corresponding to the priority, and the band is adjusted according to the number of bits to be shifted. Since the priorities of the components are determined, it is possible to efficiently specify an encoding target according to the target image quality of the compressed image, and perform high-speed code amount control with a small amount of calculation.
[0139]
According to the second and tenth aspects, it is possible to control the code amount so as to realize high display image quality suitable for human visual evaluation.
[0140]
According to the third and eleventh aspects, the code amount can be finely and efficiently controlled in bit plane units according to the target image quality.
[0141]
According to the fourth and twelfth aspects, it is possible to finely and efficiently control the code amount for each coding pass according to the target image quality.
[0142]
According to the fifth and sixth aspects and the thirteenth and fourteenth aspects, the code amount control of the compressed encoded data is efficiently performed according to the priority set for each band component without decoding the compressed encoded data. It is possible. Further, even without using the rate / distortion optimization, it is possible to perform code amount control with high real-time property so that distortion can be suppressed.
[0143]
According to the seventh and fifteenth aspects, the band component bit-shifted according to the priority is divided into a plurality of layers. Therefore, it is possible to efficiently generate a plurality of layers so that distortion with respect to a rate can be reduced. It is possible.
[0144]
According to the eighth and sixteenth aspects, it is possible to generate a compressed image having high display quality and suitable for human visual evaluation.
[Brief description of the drawings]
FIG. 1 is a functional block diagram illustrating a schematic configuration of a code amount control device according to an embodiment of the present invention.
FIG. 2 is a functional block diagram showing a schematic configuration of a bit truncation control unit in the code amount control device shown in FIG.
FIG. 3 is a schematic diagram showing a two-dimensional image obtained by band division according to an octave division method.
FIG. 4 is a diagram for explaining a priority setting process by a bit shift;
FIG. 5 is a diagram illustrating an example of a bit-shifted transform coefficient;
FIG. 6 is a schematic diagram showing a two-dimensional image divided into bands by wavelet transform.
FIG. 7 is a schematic diagram showing a two-dimensional image divided into bands by wavelet transform.
FIG. 8 is a schematic diagram showing conversion coefficients of band components shifted right by bits according to the priorities shown in FIG. 7;
FIG. 9 is a diagram showing a numerical value table of Energy @ weighting @ factor.
FIG. 10 is a diagram showing a numerical value table of Energy @ weighting @ factor.
FIG. 11 is a diagram showing a numerical value table of Energy @ weighting @ factor.
FIG. 12 is a functional block diagram illustrating a schematic configuration of an image quality control unit according to the embodiment.
FIG. 13 is a schematic diagram illustrating a conversion coefficient bit-shifted according to a priority;
FIG. 14 is a diagram for describing a processing example of a transform coefficient of a band component LL5.
FIG. 15 is a diagram for describing a processing example of a transform coefficient of a band component LL5.
FIG. 16 is a diagram for describing an example of encoding processing of transform coefficients of a band component HH2.
FIG. 17 is a diagram showing a curve of rate / distortion characteristics.
FIG. 18 is a diagram illustrating an example of a scanning order.
FIG. 19 is a diagram illustrating an example of a truncation point.
FIG. 20 is a diagram illustrating a code string rearranged in bit plane units.
FIG. 21 is a diagram illustrating a code string rearranged in units of an encoding pass.
FIG. 22 is a schematic diagram illustrating encoded data divided into a plurality of layers.
FIG. 23 is a functional block diagram illustrating a schematic configuration of a compression encoding device according to the JPEG2000 system.
FIG. 24 is a schematic diagram showing a two-dimensional image divided into bands according to the octave division method.
FIG. 25 is a schematic diagram showing a two-dimensional image decomposed into a plurality of code blocks.
FIG. 26 is a schematic diagram showing a plurality of bit planes constituting a code block.
FIG. 27 is a schematic diagram showing three types of encoding passes.
[Explanation of symbols]
1 code amount control device
2. Storage device
3 @ MMU
4 bit truncation control unit
5 Multiplexer
6 Priority table
7 Layer division control unit
10 Image quality control unit
11 code amount control unit

Claims (16)

The image signal is recursively divided into high-frequency components and low-frequency components by wavelet transform, transform coefficients of a plurality of band components are calculated, and the transform coefficients are entropy-coded and the amount of code of compressed image data compressed. Code amount control device for controlling the
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and From the shifted band components, an image quality control unit that selects an encoding target in accordance with a target image quality,
A code amount control device comprising: 2. The code amount control device according to claim 1, wherein the image quality control unit has a function of using the priority weighted in consideration of human visual characteristics. 3. The code amount control device according to claim 1, wherein the image quality control unit has a function of determining the coding target in units of the bit plane. 4. 4. The code amount control device according to claim 1, wherein the image quality control unit has a function of determining the coding target in units of the coding pass. 5. The image signal is recursively divided into high-frequency components and low-frequency components by wavelet transform, transform coefficients of a plurality of band components are calculated, and the transform coefficients are entropy-coded and the amount of code of compressed image data compressed. Code amount control device for controlling the
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and From the code string generated by rearranging the shifted encoded data of the band components in a predetermined scanning order, a truncation point suitable for a target code amount is calculated, and the code string before the truncation point is output. Code amount control unit for controlling
A code amount control device comprising: The code amount control device according to claim 5, wherein
The code amount control unit generates the code string by rearranging the coded data in the order of the priorities and in the same scanning order from the high frequency side to the low frequency side at the same priority level. , Code amount control device. The code amount control device according to claim 1, wherein:
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and A layer division control unit that controls to divide the shifted band component into a plurality of layers,
The code amount control device further comprising: 8. The code amount control device according to claim 7, wherein the layer division control unit has a function of using the priority weighted in consideration of human visual characteristics. The image signal is recursively divided into high-frequency components and low-frequency components by wavelet transform, transform coefficients of a plurality of band components are calculated, and the transform coefficients are entropy-coded and the amount of code of compressed image data compressed. A program for controlling the
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and From the shifted band components, as an image quality control unit that selects an encoding target according to a target image quality,
A program for causing a microprocessor to function. The program according to claim 9, wherein the image quality control unit causes the microprocessor to function so as to use the priority weighted in consideration of human visual characteristics. The program according to claim 9, wherein the image quality control unit causes the microprocessor to function so as to determine the encoding target in units of the bit plane. The program according to any one of claims 9 to 11, wherein the image quality control unit causes the microprocessor to function so as to determine the encoding target in units of the encoding pass. The image signal is recursively divided into high-frequency components and low-frequency components by wavelet transform, transform coefficients of a plurality of band components are calculated, and the transform coefficients are entropy-coded and the amount of code of compressed image data compressed. A program for controlling the
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and From the code string generated by rearranging the shifted encoded data of the band components in a predetermined scanning order, a truncation point suitable for a target code amount is calculated, and the code string before the truncation point is output. As a code amount control unit that controls
A program for causing a microprocessor to function. 14. The program according to claim 13, wherein
The code amount control unit generates the code string by rearranging the coded data in the order of the priorities and in the same scanning order from the high frequency side to the low frequency side at the same priority level. That causes the microprocessor to function. A program according to any one of claims 9 to 14, wherein
For each of the band components of the compressed image data, the band component is bit-shifted by the number of bits corresponding to the priority set according to the number of times the band is recursively divided into the low band components, and As a layer division control unit that controls to divide the shifted band component into a plurality of layers,
A program that causes the microprocessor to function. 16. The program according to claim 15, wherein the layer division control unit causes the microprocessor to function so as to use the priority weighted in consideration of human visual characteristics.

Images