|
« click pentru a vizualiza toate articolele de la aceasta categorie |
|
|
Adaugat in data de 18-09-2015 |
Carrefour Romania lanseaza colectia de toamna 2015 si website-ul tex.carrefour.ro
Alte stiri la aceasta categorie
Carrefour România lansează noua colectie de toamna, ce propune tinute moderne, confortabile si de calitate, la preturi avantajoase, disponibila acum si pe noul site: tex.carrefour.ro.
Colectia de toamna 2015 vine cu oferte atragatoare de piese vestimentare, propunand o gamă variată de stiluri în tendintele acestei toamne: de la smart-casual, urban sau graphic pana la rock si retro. Cromatica acestui sezon se concentreaza pe tonuri neutre de alb, negru, gri sau albastru-bleumarin, care se imbina cu nuante aprinse de bordeaux, violet pruna si albastru electric, iar materialele de calitate cu texturi robuste precum denimul sau pielea ecologica complementează modelele fluide, pentru un stil mereu la modă.
Colectia continuă croielile drepte si liniile clasice, pentru un stil office, dar ofera si combinatii versatile, pentru tinute nonconformiste celor cu spiritul tanar: cămăsi si rochii în printuri sau carouri, blugi clasici sau skinny, pulovere si jachete tricotate, alaturi de botine retro si accesorii sic, in culorile sezonului.
„ Colectia de toamna 2015 continua directia cu care ne-am obisnuit deja clientii, oferind tinute cu atitudine si stil, moderne, confortabile si de calitate, la preturi accesibile tuturor. Si pentru a veni si mai aproape de ei, le punem la dispozitie o varianta usoara si interactiva de a fi in pas cu tendintele prin lansarea site-ului tex.carrefour.ro. Va invitam sa va alegeti tinutele preferate, pentru a da culoare si caracter acestei toamne.” a declarat Andreea Mihai, director de marketing al Carrefour România.
Mai mult, pentru a marca lansarea noii colectii, sambata, 19 septembrie, hipermarketurile Carrefour invita clientii sa participe la o Super Tombola unde pot castiga, din 30 in 30 de minute, 50% din valoarea unui articol vestimentar din cosul de cumparaturi.
In cadrul colectiei, gama TeX cuprinde:
TeX Basic - tinute de zi, comode, ce cuprind pantaloni, tricouri, cămăsi, jachete
TeX Casual - o categorie dedicată persoanelor care caută confortul, în tinute comode
TeX Sport – articole vestimentare sport, confectionate din materiale de calitate superioară: tesături anti-perspirante si rezistente la uzură
TeX Lenjerie - gama de lenjerie care se adresează atât femeilor si bărbatilor, cât si copiilor, ce include modele atractive, realizate din materiale de o calitate superioară
TeX Nightwear (pijamale) - îmbrăcăminte pentru noapte, din materiale textile plăcute si confortabile
Pentru un plus de inspiratie, vă invităm în magazinele noastre, sa incercati articolele vestimentare sau sa descoperiti mai multe tinute din Colectia de Toamna 2015, direct pe noul site tex.carrefour.ro
Cu peste 10.100 de magazine în 34 de tări, Grupul Carrefour este al doilea retailer mondial si numărul unu în Europa. De peste 50 de ani, Carrefour este un partener pentru viata de zi cu zi. În fiecare zi, acesta primeste în magazinele sale peste 10 milioane de clienti din lumea întreagă, oferindu-le o gamă largă de produse si servicii la preturi echitabile.
În România, Grupul Carrefour numără 180 de magazine dintre care 28 hipermarketuri 'Carrefour', 98 supermarketuri 'Market', 43 magazine de proximitate 'Express', 10 magazine de proximitate 'Contact' si un magazin de comert online: www.carrefour-online.ro..
|
|
» Adauga comentariul tau |
Comentarii |
|
|
Rony
1. ä»‹ç» ä¼ ç»Ÿä¿¡å·æˆ–者图åƒé‡‡æ ·å¤šé‡‡å–é¦™å†œé‡‡æ ·å®šç†ï¼šå³é‡‡æ ·é¢‘率为信å·é¢‘çŽ‡æœ€å¤§å€¼ï¼ˆé‚£å¥Žæ–¯ç‰¹é‡‡æ ·é¢‘çŽ‡ï¼‰çš„äºŒå€ã€‚本文æ出一ç§æ–°çš„ä¿¡å·å¤„ç†æ€è·¯â€”â€”åŽ‹ç¼©é‡‡æ ·ï¼ˆCS)。在这ç§æ–°æ–¹æ³•ä¸‹ï¼Œé‡‡æ ·çŽ‡ä¼šå¤§å¤§é™ä½Žã€‚åŽ‹ç¼©é‡‡æ ·ä¾èµ–两æ¡åŽŸåˆ™ï¼šç¨€ç–性和ä¸ç›¸å¹²æ€§ï¼Œå‰è€…从属于信å·ï¼ŒåŽè€…从属于感测模å¼ã€‚稀ç–性(sparsity):表达的概念是连ç»æ—¶é—´ä¿¡å·çš„ä¿¡æ¯é‡‡é›†çŽ‡å¯èƒ½è¦æ¯”按带宽选择的采集率å°å¾—多,或者说离散时间信å·ä¾èµ–的自由度数è¦æ¯”他的长度å°çš„å¤šï¼Œæ›´æ˜Žç¡®çš„çš„è¯´ï¼ŒåŽ‹ç¼©é‡‡æ ·æŽ¢ç´¢å¦‚ä¸‹äº‹å®žï¼šè®¸å¤šè‡ªç„¶ä¿¡å·åœ¨ä¸‹è¿°æ„义下是稀ç–的或å¯åŽ‹ç¼©çš„:当用适当的基表示时,它们有更简æ´çš„表达。ä¸ç›¸å¹²æ€§ï¼ˆincoherence):延伸了时间与频率之间的对å¶æ€§ï¼Œå®ƒè¡¨è¾¾çš„概念是:æ£å¦‚时间域内的Dirac和尖峰信å·åœ¨é¢‘çŽ‡åŸŸå†…æ˜¯å±•å¼€çš„é‚£æ ·ï¼Œå…·æœ‰ä»¥$latex {\\Psi}&fg=000000$稀ç–表示的对象,在获å–它们的区域内一定是展开的。æžä¸ºé‡è¦çš„是人们å¯ä»¥è®¾è®¡å‡ºæœ‰æ•ˆçš„ä¼ æ„Ÿæˆ–é‡‡æ ·æ–¹æ¡ˆï¼Œæ•æ‰åµŒå…¥åœ¨ç¨€ç–ä¿¡å·å†…有用的信æ¯å†…容,并将其压缩æˆå°‘é‡æ•°æ®ã€‚这些方法ä¸æ˜¯è‡ªé€‚应的,且åªéœ€è¦ä¸Žå°‘é‡çš„与æ供信å·ç®€æ´è¡¨è¿°ä¹‹åŸºä¸ç›¸å¹²ï¼ˆå¦åˆ™å°†å¤±åŽ»æµ‹é‡è¿‡ç¨‹å¯¹ä¿¡å·çš„ä¾èµ–性)的固定波形相关è”ï¼›æ¤å¤–,有利用数值优化由采集的少é‡æ•°æ®é‡æž„全长信å·çš„方法。æ¢è¨€ä¹‹ï¼ŒCS是一ç§éžå¸¸ç®€å•æœ‰æ•ˆçš„ä¿¡æ¯æ•èŽ·æ–¹æ³•ï¼Œå®ƒä»¥ä¸Žä¿¡å·æ— 关的方å¼å’Œä½Žé‡‡æ ·çŽ‡é‡‡æ ·ï¼Œè€ŒåŽæ ¹æ®çœ‹ä¼¼ä¸å®Œæ•´çš„测é‡æ•°æ®é›†ç”¨è®¡ç®—能力é‡æž„ä¿¡å·ã€‚2. 感知问题 对于信å·$latex {f\\left(x\\right)}&fg=000000$ï¼Œä¼ æ„Ÿæœºåˆ¶ä¸ºï¼š$latex \\displaystyle y_{k}=\\left\\langle f,\\varphi_{k}\\right\\rangle ,\\qquad k=1,\\ldots,m.\\ \\ \\ \\ \\ (1)&fg=000000$也就是说我们åªéœ€è¦å°†è¦èŽ·å¾—的对象与波形$latex {\\varphi_{k}}&fg=000000$相关è”ï¼Œè¿™æ˜¯ä¸€ä¸ªæ ‡å‡†æž¶æž„ï¼Œä¾‹å¦‚ï¼Œå¦‚æžœæ£€æµ‹æ³¢å½¢æ˜¯Dirac delta(尖峰)函数,那么$latex {y}&fg=000000$就是$latex {f}&fg=000000$åœ¨æ—¶é—´æˆ–ç©ºé—´ä¸€ä¸ªé‡‡æ ·å€¼çš„çŸ¢é‡ï¼›å¦‚果检测波形是åƒç´ 的指示函数,那么$latex {y}&fg=000000$就是数å—相机ä¸ä¼ 感器特别采集的图åƒæ•°æ®ï¼›æœ€åŽï¼Œå¦‚果检测波形是æ£å¼¦å‡½æ•°ï¼Œé‚£ä¹ˆ$latex {y}&fg=000000$就是Fourier系数,ç£å…±æŒ¯æˆåƒ (MRI)用的就是这ç§æ£€æµ‹æ¨¡æ€ã€‚尽管å¯ä»¥å»ºç«‹èµ·æ—¶ç©ºè¿žç»çš„CSç†è®ºï¼Œä½†è¿™é‡Œåªå…³å¿ƒç¦»æ•£ä¿¡å·$latex {f\\in\\mathbb{R}^{n}}&fg=000000$ï¼ŒåŽŸå› æœ‰äºŒï¼Œä¸€æ˜¯ç®€å•ï¼ŒäºŒæ˜¯çŽ°æœ‰ç†è®ºåœ¨æ¤æƒ…况下æˆç†Ÿå¾—å¤šã€‚æˆ‘ä»¬å…³å¿ƒæ¬ é‡‡æ ·çš„æƒ…å†µï¼Œå¯å¾—到的测é‡æ•°$latex {m}&fg=000000$è¿œå°äºŽä¿¡å·$latex {f}&fg=000000$çš„ç»´æ•°$latex {n}&fg=000000$。由于å„ç§åŽŸå› ,这ç§æƒ…å½¢æžä¸ºå¸¸è§ï¼Œè¬å¦‚ä¼ æ„Ÿå™¨æ•°é‡æœ‰é™ï¼Œæˆ–者æŸäº›å€ŸåŠ©äºŽä¸å散射获å–图åƒçš„花费æžä¸ºæ˜‚贵,或者是检测过程缓慢,象MRIé‚£æ ·åªèƒ½å¯¹å¯¹è±¡æ–½è¡Œå‡ 次测é‡ï¼Œç‰ç‰ã€‚这些情况产生一些é‡è¦é—®é¢˜ï¼Œæ ¹æ®$latex {m\\ll n}&fg=000000$次测é‡èƒ½å¤Ÿå‡†ç¡®çš„é‡æž„ä¿¡å·å—?能够设计出$latex {m\\ll n}&fg=000000$个检测波形æ•èŽ·åˆ°å‡ 乎所有的关于$latex {f}&fg=000000$çš„ä¿¡æ¯å—?以åŠå¦‚ä½•æ ¹æ®è¿™äº›ä¿¡æ¯é€¼è¿‘$latex {f}&fg=000000$ï¼Ÿæ¯«æ— ç–‘é—®ï¼Œè¿™äº›éƒ½æ˜¯å›°éš¾çš„äº‹æƒ…ï¼Œå› ä¸ºå¯èƒ½éœ€è¦è§£æ¬ 定线性方程组。令$latex {A}&fg=000000$是以$latex {\\varphi_{1}^{*},\\varphi_{2}^{*},\\varphi_{3}^{*}, ,\\varphi_{m}^{*}}&fg=000000$为行的$latex {m\\times n}&fg=000000$测é‡çŸ©é˜µï¼Œ$latex {\\varphi^{*}}&fg=000000$是$latex {\\varphi}&fg=000000$çš„å¤è½¬ç½®ã€‚当$latex {m0}&fg=000000$ä¸å¤§äºŽ$latex {O\\left(n^{-\\beta}\\right)}&fg=000000$,知ä¸çš„指数是5而ä¸æ˜¯4(人们相信 对æ°ä¸º$latex {\\text{\\ensuremath{\\log}}n}&fg=000000$也是æˆç«‹çš„。这è¯æ˜Žå¯ä»¥ç¨³å®šè€Œå‡†ç¡®åœ°ç”±ä¸ç›¸å¹²åŸŸä¸æˆå‰§æ€§çš„æ¬ é‡‡æ ·æ•°æ®é‡æž„近稀ç–ä¿¡å·ã€‚最åŽï¼ŒçŸ©é˜µ$latex {A=\\Phi\\Psi}&fg=000000$也å¯ä»¥æˆç«‹çº¦æŸç‰è·æ€§ï¼Œè¿™é‡Œ$latex {\\Psi}&fg=000000$是任æ„æ£äº¤åŸºï¼Œ$latex {\\Phi}&fg=000000$是从适当分布ä¸éšæœºæŠ½å–的测é‡çŸ©é˜µã€‚如果固定基$latex {\\Psi}&fg=000000$,按命题1~4æž„æˆæµ‹é‡çŸ©é˜µï¼Œå‡è‹¥ï¼š$latex \\displaystyle m\\geq C\\cdot\\log\\left(n/S\\right)\\ \\ \\ \\ \\ (15)&fg=000000$那么,$latex {A=\\Phi\\Psi}&fg=000000$将以æžå¤§çš„概率符åˆçº¦æŸç‰è·æ€§ï¼ˆRIP);å¼ä¸$latex {C}&fg=000000$为与具体情况有关的æŸä¸€å¸¸æ•°ã€‚è¿™ç§éšæœºæµ‹é‡çŸ©é˜µ$latex {\\Phi}&fg=000000$在æŸç§æ„义上是普适的,当设计测é‡ç³»ç»Ÿæ—¶ç”šè‡³ä¸éœ€çŸ¥é“稀ç–基。6. ä»€ä¹ˆæ˜¯åŽ‹ç¼©é‡‡æ ·ï¼Ÿ æ•°æ®èŽ·å–通常按下述方å¼å·¥ä½œï¼šå¤§é‡çš„æ•°æ®åªæ˜¯é‡‡é›†å‡ºæ¥ï¼Œå…¶ä¸ä¸€å¤§éƒ¨åˆ†åœ¨åŽ‹ç¼©é˜¶æ®µè¢«æŠ›å¼ƒï¼Œä¸ºäº†å‚¨å˜å’Œä¼ é€é€šå¸¸å¿…é¡»è¿™æ ·åšã€‚用本文的è¯è¯´ï¼Œäººä»¬èŽ·å¾—高分辨率的åƒç´ 数组$latex {f}&fg=000000$,计算完整的一套å˜æ¢ç³»æ•°ï¼Œå°†æœ€å¤§çš„系数编ç ,丢掉其它系数,实质上以$latex {f_{S}}&fg=000000$结æŸï¼›è¿™ç§å¤§è§„模采集数æ®ç„¶åŽåŽ‹ç¼©çš„处ç†æ–¹æ³•æ˜¯æžä¸ºæµªè´¹çš„(å¯ä»¥æƒ³è±¡ä¸€ä¸‹æ•°å—相机有百万åƒç´ 的图åƒä¼ 感器,而最终编ç 的图åƒåªæœ‰å‡ 百kå—节)。CS的处ç†æžä¸ºä¸åŒï¼Œå…¶è¡¨çŽ°å¥½åƒæ˜¯èƒ½å¤Ÿç›´æŽ¥èŽ·å–处ç†å¯¹è±¡æ°ä¸ºæœ€é‡è¦çš„ä¿¡æ¯ã€‚通过å–如第5节的$latex {O(S\\log(n/S))}&fg=000000$个éšæœºæŠ•å½±ï¼Œäººä»¬å°±æœ‰è¶³å¤Ÿçš„ä¿¡æ¯é‡æž„ä¿¡å·ï¼Œå…¶ç²¾åº¦è‡³å°‘达到$latex {f_{S}}&fg=000000$具有的精度,得到处ç†å¯¹è±¡çš„最佳$latex {S}&fg=000000$项逼近和最优压缩表示。æ¢è¨€ä¹‹ï¼ŒCSæ•°æ®èŽ·å–å议实质上是把模拟信å·è½¬æ¢æˆåŽ‹ç¼©åŽçš„æ•°å—å½¢å¼ï¼Œä½¿å¾—人们至少从原ç†ä¸Šå¯ä»¥ä»Žä»…有为数ä¸å¤šçš„ä¼ æ„Ÿå™¨å¾—åˆ°è¶…åˆ†è¾¨ä¿¡å·ã€‚采集æ¥éª¤åŽæ‰€æœ‰éœ€è¦åšçš„就是将测é‡æ•°æ®è§£åŽ‹ç¼©ã€‚料想ä¸åˆ°çš„是采集æ¥æ˜¯å›ºå®šçš„ï¼Œå°¤å…¶æ˜¯æ ¹æœ¬ä¸è¯•å›¾ç†è§£ä¿¡å·çš„结构,好åƒæ˜¯ã€Œä¸å¬å³é—»ã€ï¼ˆ hearing without listening)。CSåŒç¼–ç ç†è®ºï¼Œæ›´æ˜Žç¡®çš„说åŒReed-Solomonç¼–ç çš„ç†è®ºå’Œå®žè·µæœ‰æŸäº›è¡¨é¢ä¸Šçš„相似性。就本文讨论内容简å•åœ°è¯´ï¼Œå¦‚所周知,å¯ä»¥é‡‡ç”¨ç¼–ç ç†è®ºçš„概念如下建立CS:人们å¯ä»¥æ ¹æ®ä¿¡å·çš„å‰$latex {2S}&fg=000000$个 Fourier系数$latex \\displaystyle {y_{k}}=\\sum\\limits _{t=0}^{n-1}{{x_{t}}{e^{-i2\\pi kt/n}}},\\quad k=0,1,2,\\ldots,2S-1\\ \\ \\ \\ \\ (16)&fg=000000$或者从相继$latex {2S}&fg=000000$个信å·é¢‘率集唯一的é‡æž„任何$latex {S}&fg=000000$-稀ç–ä¿¡å·ï¼ˆå¤åŽŸä¿¡å·çš„计算æˆæœ¬å®žè´¨ä¸Šæ˜¯è§£$latex {S\\times S}&fg=000000$ Toeplitz矩阵,或计算$latex {n}&fg=000000$点FFT),这æ„味ç€å¯ä»¥ç”¨è¿™ä¸€æ–¹æ³•æµ‹é‡å¯åŽ‹ç¼©ä¿¡å·å—ï¼Ÿç”±äºŽä¸¤ä¸ªåŽŸå› ç”案是å¦å®šçš„。其一,Reed-Solomon解ç 是代数方法,ä¸èƒ½ç”¨äºŽéžç¨€ç–ä¿¡å·ï¼ˆè§£ç 是通过求多项å¼çš„æ ¹å¯»æ‰¾ä¿¡å·æ”¯æ’‘é›†ï¼‰ï¼›å…¶äºŒï¼Œæ ¹æ®ä¿¡å·çš„å‰$latex {2S}&fg=000000$个Fourier系数寻找信å·æ”¯æ’‘集的问题,甚至当信å·å…·å¤‡å‡†ç¡®ç¨€ç–性时,是éžå¸¸ä¸é€‚定的(这个问题与由少é‡å€¼é«˜åº¦é›†ä¸çš„数预测高阶多项å¼ç›¸åŒï¼‰ï¼›å“ªæ€•ç³»æ•°å—到微å°çš„扰动也会导致完全ä¸åŒçš„ç”案,这使得用有é™ç²¾åº¦çš„æ•°æ®å¯é 地计算支撑集在实际上是ä¸å¯èƒ½çš„。å之,纯粹代数方法忽略了信æ¯ç®—å的调节作用,使å–得良æ€çŸ©é˜µï¼Œè¿™æ˜¯ç²¾ç¡®ä¼°è®¡çš„关键,æ£å¦‚约æŸç‰è·æ€§æ‰€èµ·ä½œç”¨æ˜¯CS所关心的ä¸å¿ƒé—®é¢˜ã€‚7. 应用 å¯åŽ‹ç¼©ä¿¡å·å¯ä»¥ç”¨ä¸Žä¿¡æ¯æ°´å¹³$latex {S\\ll n}&fg=000000$æˆæ£æ¯”的若干ä¸ç›¸å¹²æµ‹é‡æ•èŽ·ï¼Œè¿™ä¸€äº‹å®žæœ‰ç€æ·±è¿œçš„æ„义并涉åŠåˆ°è®¸å¤šå¯èƒ½çš„应用。数æ®åŽ‹ç¼©ï¼šå¯¹æ•°æ®åŽ‹ç¼©è€Œè¨€ï¼Œåœ¨æŸäº›æƒ…况下解ç 器上稀ç–基å¯èƒ½æ˜¯æœªçŸ¥çš„,或者施行起æ¥ä¸å®žç”¨ã€‚然而æ£å¦‚第五节所讨论的éšæœºè®¾è®¡çš„$latex {\\Phi}&fg=000000$å¯è§†ä¸ºæ™®é€‚解ç ç–ç•¥ï¼Œå› ä¸ºå®ƒä¸éœ€è¦å›´ç»•$latex {\\Psi}&fg=000000$的结构设计(åªæœ‰è§£ç 或å¤åŽŸ$latex {f}&fg=000000$æ—¶æ‰éœ€å…·å¤‡$latex {\\Psi}&fg=000000$的知识和实现$latex {\\Psi}&fg=000000$的能力)。这ç§æ™®é€‚æ€§ç‰¹åˆ«æœ‰åŠ©äºŽè¯¸å¦‚ä¼ æ„Ÿå™¨ç½‘ç»œä¹‹ç±»çš„å¤šä¿¡å·è£…置的编ç 。关于这个问题请读者å‚考Nowakç‰äºº åŠGoyal在其它地方的论文。信é“ç¼–ç :æ£å¦‚文献{[}15{]}æ‰€è§£é‡Šçš„é‚£æ ·ï¼Œå¯ä»¥å›´ç»•CS的原ç†ï¼ˆç¨€ç–性,éšæœºæ€§ï¼Œå’Œå‡¸ä¼˜åŒ–ï¼‰å¹¶å°†å…¶ç”¨äºŽè®¾è®¡å¿«é€Ÿçº é”™ç ,é¿å…错误信å·çš„ä¼ è¾“ã€‚å问题:对于其它一些情况,获å–$latex {f}&fg=000000$的唯一方法å¯èƒ½æ˜¯ç”¨æŸç§æ¨¡æ€çš„测é‡ç³»ç»Ÿ$latex {\\Phi}&fg=000000$,å‡å®šä¸Ž$latex {\\Phi}&fg=000000$ä¸ç›¸å¹²$latex {f}&fg=000000$的稀ç–基$latex {\\Psi}&fg=000000$å˜åœ¨ã€‚就有å¯èƒ½è¿›è¡Œæœ‰æ•ˆçš„测é‡ï¼Œä¸€ä¸ªåº”用涉åŠåˆ°MRè¡€ç®¡é€ å½±æœ¯å’Œå…¶å®ƒç±»åž‹çš„MR设备;在这些例åä¸$latex {\\Phi}&fg=000000$记录$latex {f}&fg=000000$çš„Fourierå˜æ¢çš„å集,所希望的图åƒ$latex {f}&fg=000000$在时间域和å°æ³¢åŸŸéƒ½æ˜¯ç¨€ç–的。其它领域的这个问题Lustigç‰äººæœ‰æ›´æ·±å…¥çš„讨论。数æ®é‡‡é›†ï¼šæœ‰æ—¶å…¨éƒ¨æµ‹é‡æ¨¡æ‹Ÿä¿¡å·çš„nä¸ªç¦»æ•£æ—¶é—´æ ·æœ¬å¯èƒ½éš¾ä»¥å¾—到(而且接下æ¥ä¹Ÿéš¾ä»¥åŽ‹ç¼©ï¼‰ã€‚在这ç§æƒ…况下,å¯èƒ½æœ‰åŠ©äºŽè®¾è®¡å‡ºç‰©ç†é‡‡æ ·è®¾å¤‡ï¼Œç›´æŽ¥è®°å½•ä¼ 入模拟信å·ç¦»æ•£çš„ã€ä½Žé‡‡æ ·çŽ‡çš„ä¸ç›¸å¹²æµ‹é‡ã€‚最åŽä¸€ä¸ªåº”用暗示我们,数å¦å’Œè®¡ç®—方法å¯èƒ½ä¼šå¯¹ä¼ 统的硬件设计有é™åˆ¶çš„é¢†åŸŸäº§ç”Ÿå·¨å¤§çš„å†²å‡»ã€‚å¦‚ä¼ ç»Ÿçš„å›¾åƒè®¾å¤‡é‡‡ç”¨CCDå’ŒCMOS技术基本上é™äºŽå¯è§†å…‰è°±ï¼Œè€ŒCS相机用数å—微型镜头阵列采集ä¸ç›¸å¹²æ•°æ®ï¼Œä¹Ÿè®¸èƒ½å¤Ÿæ˜Žæ˜¾æ‰©å……其能力。关于这个问题其它地方的讨论,以Baraniuk讨论这ç§ç›¸æœºæ›´åŠ æ·±å…¥ã€‚æœ‰ä¸€éƒ¨åˆ†ç ”ç©¶è€…å·²ä¸“å¿ƒäºŽå¤§å¸¦å®½é«˜çº§æ¨¡æ‹Ÿ-ä¿¡æ¯ï¼ˆA/I)转æ¢è®¾å¤‡çš„ç ”ç©¶ï¼Œç›®æ ‡æ˜¯å¸®åŠ©å‡è½»ä¼ 统模拟-æ•°å—(A/D)转æ¢æŠ€æœ¯çš„压力,当å‰è¯¥æŠ€æœ¯é™äºŽ1GHzçš„é‡‡æ ·é€Ÿåº¦ã€‚ä½œä¸ºå¯é€‰æ‹©çš„方案,我们建议两ç§ç‰¹æ®Šçš„ã€å¯ä»¥ä»Žé«˜å¸¦å®½æ¨¡æ‹Ÿä¿¡å·ä¸èŽ·å¾—离散的ã€ä½Žé‡‡æ ·çŽ‡ï¼Œä¸ç›¸å¹²æµ‹é‡åºåˆ—çš„A/I架构。对高阶逼近,æ¯ä¸€ä¸ªæµ‹é‡$latex {y_{k}}&fg=000000$,都å¯ä»¥è§£é‡Šä¸ºå…¥å°„模拟信å·$latex {f}&fg=000000$与模拟测é‡æ³¢å½¢$latex {\\varphi_{k}}&fg=000000$的内积$latex {\\left\\langle f,\\varphi_{k}\\right\\rangle }&fg=000000$,æ£å¦‚在离散CSæ¡†æž¶é‚£æ ·ï¼Œæˆ‘ä»¬çš„åˆæ¥ç»“果表明éµå¾ªç¨€ç–或å¯åŽ‹ç¼©æ¨¡åž‹ï¼ˆåœ¨æŸä¸€æ¨¡æ‹Ÿå—å…¸$latex {\\Psi}&fg=000000$ä¸ï¼‰çš„ä¿¡å·å¯ä»¥ç”¨è¿™äº›è®¾å¤‡ä»¥æ£æ¯”于信æ¯æ°´å¹³è€Œä¸æ˜¯é‚£å¥Žæ–¯ç‰¹é¢‘率有效的æ•èŽ·ã€‚当然,利用离散CS方法å¤åŽŸæ¨¡æ‹Ÿç¨€ç–ä¿¡å·æœ‰å¾…å…‹æœçš„挑战性问题。全é¢è§£å†³è¿™äº›é—®é¢˜è¶…出了本çŸæ–‡çš„范围;人们å¯ä»¥ç®€å•çš„接å—å…¶æ€æƒ³ï¼Œç¦»æ•£/é‡‡æ ·ç¨€ç–å—å…¸å…许适当的å¤åŽŸã€‚我们有两个架构如下:}~éžå‡åŒ€é‡‡æ ·å™¨ï¼ˆNUS):æ¤æž¶æž„åªæ˜¯æŠŠéšæœºé‡‡æ ·æ—¶é—´ç‚¹ä¸Šçš„ä¿¡å·æ•°å—化,å³$latex {{y_{k}}=f\\left({t_{k}}\\right)=\\left\\langle {f,{\\delta_{{t_{k}}}}}\\right\\rangle }&fg=000000$,事实上这些éšæœºæˆ–伪éšæœºæ—¶é—´ç‚¹æ˜¯é€šè¿‡æŠ–åŠ¨è§„åˆ™æ ¼å上的åä¹‰ï¼ˆä½Žé‡‡æ ·çŽ‡ï¼‰é‡‡æ ·ç‚¹å¾—åˆ°çš„ã€‚ç”±äºŽå°–å³°ä¸Žæ£å¼¦ä¿¡å·ä¸ç›¸å¹²ï¼Œå› æ¤è¿™ç§æž¶æž„å¯ç”¨äºŽå¯¹å…·æœ‰è¿œä½ŽäºŽé‚£å¥Žæ–¯ç‰¹é¢‘率的稀ç–频谱信å·é‡‡æ ·ã€‚当然与å‡å°‘é‡‡æ ·çŽ‡æœ‰å…³çš„å¥½å¤„æ˜¯æžå¤§çš„ï¼Œå› ä¸ºè¿™æä¾›äº†é™„åŠ ç”µè·¯ç¨³å®šæ—¶é—´ï¼Œå¹¶å…·æœ‰å‡å°å™ªå£°æ°´å¹³çš„作用。éšæœºé¢„积分(RPI):æ¤æž¶æž„å¯ç”¨äºŽç§ç±»æ›´å¹¿æ³›çš„稀ç–域,在时间-频率平é¢æœ‰ç¨€ç–特å¾çš„最显著的那些信å·ã€‚鉴于没有å¯èƒ½ä»¥æžé«˜çš„é‡‡æ ·çŽ‡å°†æ¨¡æ‹Ÿä¿¡å·æ•°å—化,而æžæœ‰å¯èƒ½ä»¥å¾ˆé«˜çš„速率改å˜å®ƒçš„æžæ€§ã€‚RPI架构的æ€æƒ³ï¼ˆè§å›¾5)就是以æ£1和负1的伪éšæœºåºåˆ—乘以信å·ï¼Œå°†ä¹˜ç§¯åœ¨æ—¶é—´çª—å£ç§¯åˆ†ï¼Œå°†æ—¶æ®µæœ«çš„积分值数å—åŒ–ï¼Œè¿™æ˜¯ä¸€ä¸ªå¹¶è¡Œæž¶æž„ï¼Œæœ‰è‹¥å¹²ä¸ªè¿™æ ·çš„ä¹˜æ³•å™¨-积分器对用完全ä¸åŒçš„ç”šè‡³å‡ ä¹Žç‹¬ç«‹çš„éšæœºç¬¦å·åºåˆ—并列è¿è¡Œã€‚事实上RPI架构将信å·åŒ$latex {\\pm1}&fg=000000$符å·åºåˆ—库相关è”,是普适的CS测é‡æ–¹æ³•ä¹‹ä¸€ã€‚对上述æ¯ä¸ªæž¶æž„,我们都曾用数值的方法确认,有些还用物ç†çš„方法è¯å®žï¼Œç³»ç»Ÿå¯¹äºŽç”µè·¯éžç†æƒ³ï¼Œå¦‚çƒå™ªå£°ã€æ—¶é’Ÿè¯¯å·®ã€å¹²æ‰°åŠæ”¾å¤§å™¨éžçº¿æ€§ç‰æƒ…况是é²æ£’性的。将A/I架构应用于实际采集方案还需è¦å‘展CS算法和ç†è®ºï¼ŒåŒ…æ‹¬ç ”ç©¶æ¨¡æ‹Ÿä¿¡å·ç¨€ç–表示的å—典。我们以最åŽä¸€ä¸ªç¦»æ•£çš„例å结æŸã€‚对这个实验,我们å–$latex {f}&fg=000000$是长度$latex {n=512}&fg=000000$çš„1-Dä¿¡å·ï¼ŒåŒ…å«ä¸¤ä¸ªè°ƒåˆ¶è„‰å†²ï¼ˆå›¾6左侧),从这个信å·ä¸æˆ‘们用以独立åŒåˆ†å¸ƒæŸåŠªåˆ©$latex {\\pm1}&fg=000000$å…ƒç´ å¡«å……ç”Ÿæˆçš„$latex {m\\times n}&fg=000000$测é‡çŸ©é˜µ$latex {\\Phi}&fg=000000$, 采集$latex {m=30}&fg=000000$次测é‡ï¼Œè¿™æ˜¯ä¸åˆç†çš„å°æ•°æ®é‡ï¼Œæ¬ é‡‡æ ·å› å超过17。为了é‡æž„ä¿¡å·ï¼Œæˆ‘们考虑时-频Gaborå—å…¸$latex {\\Psi}&fg=000000$,它由Gaussian窗é™æ—¶å¹¶å…·æœ‰ä¸åŒä½ç½®å’Œå°ºåº¦çš„å„ç§æ£å¼¦æ³¢æž„æˆã€‚总的说å—典是过完备($latex {43\\times}&fg=000000$overcomplete)的,ä¸å«åŒ…括$latex {f}&fg=000000$的两个脉冲。图6ä¸é—´å›¾å½¢è¡¨ç¤ºå–$latex {{\\left\\Vert x\\right\\Vert _{{\\ell_{1}}}}}&fg=000000$æžå°åŒ–使$latex {y=\\Phi\\Psi x}&fg=000000$,é‡æž„效果显示出明显的人为效应,我们看到$latex {{\\left\\Vert {f-{f^{*}}}\\right\\Vert _{{\\ell_{2}}}}/{\\left\\Vert f\\right\\Vert _{{\\ell_{2}}}}\\approx0.67}&fg=000000$,而事实上我们通过将两者都å˜ä¸º$latex {\\ell_{1}}&fg=000000$å¤åŽŸæ–¹æ³•ï¼Œå°±å¯ä»¥æ¶ˆé™¤äººä¸ºæ•ˆåº”。首先,我们改用æžå°åŒ–$latex {{\\left\\Vert {\\Psi*\\tilde{f}}\\right\\Vert _{{\\ell_{1}}}}\\; s.t.\\; y=\\Phi\\tilde{f}}&fg=000000$;(当$latex {\\Psi}&fg=000000$是æ£äº¤åŸºæ—¶è¿™ä¸€æ”¹å˜ä¸èµ·ä½œç”¨ï¼‰ï¼›å…¶æ¬¡ï¼Œå½“得到估计值$latex {f^{*}}&fg=000000$时我们将$latex {\\ell_{1}}&fg=000000$-范数é‡æ–°åŠ æƒï¼Œé‡å¤é‡æž„过程,对那些估计较大的系数采用较低的惩罚值;图6å³å›¾æ˜¾ç¤ºäº†é‡å¤åŠ æƒ4次è¿ä»£çš„结果,我们看到$latex {{\\left\\Vert {f-{f^{*}}}\\right\\Vert _{{\\ell_{2}}}}/{\\left\\Vert f\\right\\Vert _{{\\ell_{2}}}}\\approx0.022}&fg=000000$。有关这个新方å‘的更多信æ¯ï¼Œå»ºè®®è¯»è€…å‚阅文献{[}30{]}。在æ¤ï¼Œç»™å‡ºçš„è¦ç‚¹æ˜¯ï¼Œå³ä¾¿è¿™ä¸ªæ•°æ®é‡å°åˆ°è’谬的地æ¥ï¼Œäººä»¬ä»å¯ä»¥æ•èŽ·åˆ°ä¿¡å·é‡ŒåŒ…å«çš„ç»å¤§éƒ¨åˆ†ä¿¡æ¯ã€‚这就是CS在当å‰ä»¥åŠæœªæ¥åº”用æžæœ‰å‰é€”çš„åŽŸå› ã€‚a0References[1] E.J. Cand`es, J. Romberg, and T. Tao. Robust uncertainty principles: Exact signal reconstructionfrom highly incomplete frequency information. Information Theory, IEEE Transactionson, 52(2):489–509, 2006.[2] E.J. Candes and T. Tao. Near-optimal signal recovery from random projections: Universalencoding strategies? Information Theory, IEEE Transactions on, 52(12):5406–5425, 2006.[3] D.L. Donoho. Compressed sensing. Information Theory, IEEE Transactions on,52(4):1289–1306, 2006.[4] A. Bilgin, M.W. Marcellin, and M.I. Altbach. Compression of electrocardiogram signals usingJPEG2000. Consumer Electronics, IEEE Transactions on, 49(4):833–840, 2003.[5] D.L. Donoho and X. Huo. Uncertainty principles and ideal atomic decomposition. InformationTheory, IEEE Transactions on, 47(7):2845–2862, 2001.[6] R. Coifman, F. Geshwind, and Y. Meyer. Noiselets. Applied and Computational HarmonicAnalysis, 10(1):27–44, 2001.[7] J.F. Claerbout and F. Muir. Robust modeling with erratic data. Geophysics, 38:826, 1973.[8] F. Santosa and W.W. Symes. Linear inversion of band-limited reflection seismograms. SIAMJournal on Scientific and Statistical Computing, 7:1307, 1986.[9] J. Tropp and A.C. Gilbert. Signal recovery from partial information via orthogonal matchingpursuit, 2005.[10] E. Cand`es and J. Romberg. Sparsity and incoherence in compressive sampling. Inverseproblems, 23:969, 2007.[11] A.C. Gilbert, S. Muthukrishnan, and M. Strauss. Improved time bounds for near-optimalsparse Fourier representations. In Proceedings of SPIE, volume 5914, page 59141A. Citeseer,2005.[12] M. Vetterli, P. Marziliano, and T. Blu. Sampling signals with finite rate of innovation. SignalProcessing, IEEE Transactions on, 50(6):1417–1428, 2002.[13] D.L. Donoho and P.B. Stark. Uncertainty principles and signal recovery. SIAM Journal onApplied Mathematics, 49(3):906–931, 1989.[14] P. Feng and Y. Bresler. Spectrum-blind minimum-rate sampling and reconstruction of multibandsignals. In icassp, pages 1688–1691. IEEE, 1996.[15] E.J. Candes and T. Tao. Decoding by linear programming. Information Theory, IEEETransactions on, 51(12):4203–4215, 2005.[16] E.J. Candes, J.K. Romberg, and T. Tao. Stable signal recovery from incomplete and inaccuratemeasurements. Communications on Pure and Applied Mathematics, 59(8):1207–1223,2006.[17] EJ Cand`es. Lectures on compressive sampling and frontiers in signal processing. The Institutefor Mathematics and its Applications, pages 2006–2007.[18] A. Cohen, W. Dahmen, and R. DeVore. Compressed sensing and best k-term approximation.American Mathematical Society, 22(1):211–231, 2009.[19] E. Candes and T. Tao. The Dantzig selector: Statistical estimation when p is much largerthan n. The Annals of Statistics, 35(6):2313–2351, 2007.[20] J. Haupt and R. Nowak. Signal reconstruction from noisy random projections. InformationTheory, IEEE Transactions on, 52(9):4036–4048, 2006.[21] R. Tibshirani. Regression shrinkage and selection via the lasso. Journal of the Royal StatisticalSociety. Series B (Methodological), 58(1):267–288, 1996.[22] S.S. Chen, D.L. Donoho, and M.A. Saunders. Atomic decomposition by basis pursuit. SIAMjournal on scientific computing, 20(1):33–61, 1999.[23] R. Baraniuk, M. Davenport, R. DeVore, and M. Wakin. A simple proof of the restrictedisometry proetrpy for random matrices. Constructive Approximation, 28(3):253–263, 2008.[24] S. Mendelson, A. Pajor, and N. Tomczak-Jaegermann. Uniform uncertainty principle forBernoulli and subgaussian ensembles. Constructive Approximation, 28(3):277–289, 2008.[25] M. Rudelson and R. Vershynin. On sparse reconstruction from Fourier and Gaussian measurements.Communications on Pure and Applied Mathematics, 61(8):1025–1045, 2008.[26] R.E. Blahut. Algebraic codes for data transmission. Cambridge Univ Pr, 2003.[27] D. Baron, M.B. Wakin, M.F. Duarte, S. Sarvotham, and R.G. Baraniuk. Distributed compressedsensing. preprint, pages 1–50, 2005.[28] M. Lustig, D.L. Donoho, and J.M. Pauly. Rapid MR imaging with compressed sensing andrandomly under-sampled 3DFT trajectories. In Proc. 14th Ann. Meeting ISMRM. Citeseer.[29] D. Takhar, V. Bansal, M. Wakin, M. Duarte, D. Baron, KF Kelly, and RG Baraniuk. Acompressed sensing camera: New theory and an implementation using digital micromirrors.Proc. Computational Imaging IV at SPIE Electronic Imaging, San Jose, 2006.[30] E.J. Candes, M.B. Wakin, and S.P. Boyd. Enhancing sparsity by reweighted l1 minimization.Journal of Fourier Analysis and Applications, 14(5):877–905, 2008.
|
« click pentru a vizualiza toate articolele de la aceasta categorie |
|
Recomanda aceasta pagina |
Daca considerati acest articol interesant il puteti recomanda unui prieten folosind formularul alaturat. |
|
|
|
|