2015/5/22 1 ã¤ã³ã¿ã©ã¯ãã£ãã·ã¹ãã è« ç¬¬ï¼å 梶æ¬è£ä¹ Twitter ID kajimoto ããã·ã¥ã¿ã° #ninshiki ã¬ãã¼ãã¡ã¼ã«ã¨ã©ã¼ 第ï¼ååã®ã¬ãã¼ãã¡ã¼ã«ãã¨ã©ã¼ã« ãªã£ãå¯è½æ§ãããã¾ãï¼ï¼ã¡ã¼ã« ããã¯ã¹ã®ãµã¤ãºãªã¼ãã¼ï¼ å度éä¿¡ãã¦ãã ããï¼ æ¥ç¨ 4/9 ã¤ã³ãããã¯ã·ã§ã³ 4/16 Scilabã®ç´¹ä»ï¼è¥¿6å·é¤¨3éPCã«ã¼ã ï¼ 4/23 åºå¼µã«ããä¼è¬ 4/30 ãã¼ãªã¨å¤æ 5/7 ãã¼ãªã¨å¤æã¨ç·å½¢ã·ã¹ãã 5/14 ä¿¡å·å¦çã®åºç¤ 5/21 ä¿¡å·å¦çå¿ç¨ï¼ï¼ç¸é¢ï¼ 5/28 ä¿¡å·å¦çå¿ç¨ï¼ï¼ç»åå¦ç) 6/4 ç 究室è¦å¦ï¼å¤§å¦é¢ãªã¼ãã³ã©ãï¼ 6/11 ä¸é確èªãã¹ã 6/18 ã©ãã©ã¹å¤æ 6/25 åºå¼µã«ããä¼è¬ 7/2 å¤å ¸å¶å¾¡ã®åºç¤ 7/9 è¡å 7/16 è¡åã¨æå°äºä¹æ³ 7/23 ãããã£ã¯ã¹ 8/5ï½11 ææ«ãã¹ã ä¸é
ã«ã¡ã©ãã£ãªãã¬ã¼ã·ã§ã³ã¨3次å åæ§æ¶ ãã®ã»ã¯ã·ã§ã³ã§è¿°ã¹ãé¢æ°ã¯ï¼ãããããã³ãã¼ã«ã«ã¡ã©ã¢ãã«ãåãæ±ãã¾ãï¼ã¤ã¾ããã®ã¢ãã«ã§ã¯ï¼3次å 座ç¹ãéè¦æå½±å¤æãç¨ãã¦ç»åå¹³é¢ã«å°å½±ãããã¨ã§ï¼ã·ã¼ã³ã®ãã¥ã¼ãæ§æããã¦ãã¾ãï¼ ãã㧠ã¯ã¯ã¼ã«ã座æ¨ç³»ã®3次å 座æ¨ã表ã ã¯ç»åå¹³é¢ã«æå½±ãããç¹ã®åº§æ¨ã表ãã¾ãï¼ ã¯ï¼ã«ã¡ã©è¡åï¼ã¾ãã¯ã«ã¡ã©ã®å é¨ãã©ã¡ã¼ã¿è¡åã¨å¼ã°ãã¾ãï¼ ã¯ä¸»ç¹ï¼é常ã¯ç»åä¸å¿ï¼ï¼ ã¯ãã¯ã»ã«åä½ã§è¡¨ãããç¦ç¹è·é¢ã§ãï¼ ãããã£ã¦ï¼ ãããã¡ã¯ã¿ã«ãã£ã¦ã«ã¡ã©ç»åãã¹ã±ã¼ãªã³ã°ããã¦ããå ´åï¼ ãã®ãã¹ã¦ã®ãã©ã¡ã¼ã¿ãåããã¡ã¯ã¿ã§ã¹ã±ã¼ãªã³ã°ï¼ãããããï¼æ¡å¤§ã¾ãã¯ç¸®å°ï¼ããå¿ è¦ãããã¾ãï¼ å é¨ãã©ã¡ã¼ã¿è¡åã¯ãã¥ã¼ã«ä¾åããªãã®ã§ï¼ä¸åº¦æ¨å®ããã°ï¼ãºã¼ã ã¬ã³ãºã®å ´åï¼ç¦ç¹è·é¢ãåºå®ãã¦ããéãã¯ç¹°è¿ã使ç¨ãããã¨ãã§ãã¾ãï¼ ä¸¦é²-å転ã®å次å¤æè¡åã§ã
ãã®ä¾ã§ã¯ãHOG ç¹å¾´ããã³ãã«ãã¯ã©ã¹ SVM åé¡å¨ã使ç¨ãã¦æ°åãåé¡ããæ¹æ³ã説æãã¾ãã ãªãã¸ã§ã¯ãã®åé¡ã¯ãç£è¦ãèªåè»å®å ¨æ§ãã¤ã¡ã¼ã¸ã®æ¤ç´¢ãªã©ãå¤ãã®ã³ã³ãã¥ã¼ã¿ã¼ ãã¸ã§ã³ ã¢ããªã±ã¼ã·ã§ã³ã«ããã¦éè¦ãªã¿ã¹ã¯ã§ãããã¨ãã°ãèªåè»å®å ¨æ§ã¢ããªã±ã¼ã·ã§ã³ã§ã¯ãè¿ãã®ãªãã¸ã§ã¯ããæ©è¡è ãè»ä¸¡ã¨ãã¦åé¡ããªããã°ãªããªããã¨ãããã¾ããåé¡ãããªãã¸ã§ã¯ãã®ã¿ã¤ãã«ãããããããªãã¸ã§ã¯ãåé¡å¨ãä½æããåºæ¬çãªæé ã¯æ¬¡ã®ããã«ãªãã¾ãã 対象ãªãã¸ã§ã¯ãã®ã¤ã¡ã¼ã¸ãå«ãã©ãã«ä»ããã¼ã¿ã»ãããåå¾ããã ãã¼ã¿ã»ãããå¦ç¿ã»ããã¨ãã¹ã ã»ããã«åå²ããã å¦ç¿ã»ããããæ½åºãããç¹å¾´ã使ç¨ãã¦åé¡å¨ã«å¦ç¿ãè¡ãããã ãã¹ã ã»ããããæ½åºãããç¹å¾´ã使ç¨ãã¦åé¡å¨ããã¹ãããã 説æã®ããã«ããã®ä¾ã§ã¯ HOG (å¾é æ¹åãã¹ãã°ã©ã ) ç¹å¾´ [1] ã¨ãã«ãã¯ã©
注ç®ãã¦ããç»ç´ ã¨ãã®è¿åã®ç»ç´ ã®æ¿åº¦å¤ã«ãããéã¿ä»ãããããã¨ããããã®åãã¨ã£ã¦ã注ç®ãã¦ããç»ç´ ã®æ°ããæ¿åº¦å¤ã¨ãããããªå¦çãè¿åå¦çã¨ããã¾ãããã®ã¨ããéã¿ä»ãã«ç¨ããå¤ã¯ããªãã¬ã¼ã¿ãªã©ã¨å¼ã°ãã¾ããä»åã¯ã3Ã3 ã®è¿åé åãç¨ãã空éãã£ã«ã¿ã«ã¤ãã¦èãã¦ãããã¨ã«ãã¾ãã 注ç®ãã¦ããç»ç´ ã®æ¿åº¦å¤ã f (i , j ) ã¨ããã¨ãå¦ç対象ã¨ãªãæ¿åº¦å¤ã¯ã表.2(a)ã®ããã«ãªãã¾ããã¾ãããªãã¬ã¼ã¿ a (k , l ) ã¯è¡¨.2(b)ã«ç¤ºãã¾ãã
ã¯ããã¾ãã¦ã æ ªå¼ä¼ç¤¾ããã¡ãã£ã¢ã§ããã³ãã¨ã³ãããã¤ã¤ãæã ãã¶ã¤ã³å¨ãã®ç£ä¿®ããã¦ãã ãã¨ã¼ ã§ãã ããã¡ãã£ã¢ã§ã¯æ¯æãæå¿ãéã¾ãåå¼·ä¼ãéãã¦ãã¾ãã ä»æ¥ã¯ããã§çºè¡¨ãããç»åå½¢å¼ã¨ãã®å 容ã«ãããå§ç¸®ã®ããããã«ã¤ãã¦ã®è³æãå ¬éãããã¨æãã¾ãã ãã®è¨äºã®å 容 ããã¼ãLPç¨ã®ç»åä½æãããã®ã®ãè¦å®å®¹éè¶ ãã¡ãã£ã! ã©ãããã... ð ã£ã¦ãããããã ãã¡ã¤ã«ã®å½¢å¼ã«ãã£ã¦ãå§ç¸®ãããããç»åã¨ããã«ããç»åãããã¾ã å§ç¸®ãããããæ§é ãç解ãããã¨ã§ã質ã®é«ã軽ãç»åãä½ãã⨠ãã®è¨äºãèªãã§ã»ãã人 容éã¨ã«ããã£ããããã¨ã®å¤ããã¶ã¤ãã¼ å°ãã§ã軽ããµã¤ããä½ãããããã³ãã¨ã³ãã¨ã³ã¸ã㢠ã¯ããã« ä¸è¬çã«ãç»åã軽ããããã¨ããã¨ãæ¸ãåºãæã®ã¯ãªãªãã£é¸æãã æ¸ãåºãå¾ã«ã¡ã¿ãã¼ã¿ãæ¤å»ãããªã©ããã¼ã«ã§å¦çããå°è±¡ãããã¨æãã¾ãã
第ï¼-ï¼åãç»åãã¡ã¤ã«ãã©ã¼ãããã(9/16,9/23) è¬ç¾©è³æ(Lec13.pdf) æ¼ç¿ããã°ã©ã (Ex07.zip) è¬ç¾©è³æ(Lec14.pdf): æ¼ç¿07ãã©ã¼ãããå¤æã®ãã³ã 第ï¼-ï¼åãå¨æ³¢æ°å解ã(10/7,10/14,10/21,10/28,11/4) è¬ç¾©è³æ(Lec15.pdf) æ¼ç¿ããã°ã©ã (Ex08.zip) è¬ç¾©è³æ(Lec16.pdf) æ¼ç¿ããã°ã©ã (Ex09.zip) 第ï¼-ï¼ï¼åããã£ã«ã¿å¦çã»ã¨ãã¸å¼·èª¿ã(11/11,11/18,11/25,12/2,12/9,12/16) è¬ç¾©è³æ(Lec17.pdf) æ¼ç¿ããã°ã©ã (Ex10.zip) è¬ç¾©è³æ(Lec18.pdf) æ¼ç¿ããã°ã©ã (Ex11.zip) è¬ç¾©è³æ(Lec19.pdf) æ¼ç¿ããã°ã©ã (Ex12.zip) è¬ç¾©è³æ(Lec20.pdf) æ¼ç¿ããã°ã©ã (Ex13.zip)
ã¡ã¤ã³ãã¼ã¸ ï¼ ç»åå¦ç äºå¤åç»åå¦çãããç»åã«ããã¦ãç½ã®é¨åï¼ã¾ãã¯é»ã®é¨åï¼ãé£ç¶ããç»ç´ ã«åãçªå·ãå²ãæ¯ãå¦çã ã©ããªã³ã°ã¨è¨ãã¾ãã é常ãåãçªå·ãã¨ã®é¢ç©ï¼ç»ç´ æ°ï¼ãå¹ ãé«ããªã©ã®ç¹å¾´éãæ±ãã¦æ¬ é¥æ¤æ»ãåé¡å¦çãªã©ã«ç¨ãããã¾ãã ã©ããªã³ã°ã«ã¯ãäºå¤åãããç»åã®ç¸¦ã横æ¹åã«é£ç¶ãã¦ããé¨åãåãã©ãã«ã«ããï¼é£çµã¨ã 縦ã横ãæãæ¹åã«é£ç¶ãã¦ããé¨åãåãã©ãã«ã«ããï¼é£çµã¨ã®ï¼ç¨®é¡ã®å¦çãããã¾ãã ï¼ï¼è¿åãï¼è¿åã¨ããå ´åãããã¾ããï¼ ä»¥ä¸ãï¼é£çµã®å ´åã«ããã¦ãã©ããªã³ã°å¦çã¢ã«ã´ãªãºã ãç´¹ä»ãã¾ãã ã¾ããç»åå ¨ã¦ã®ç»ç´ ã®ã©ãã«çªå·ãï¼ï¼ã¼ãï¼ã§åæåãã¦ãããã©ããªã³ã°ã§çªå·ãå²ãä»ããããã® ã©ããªã³ã°çªå·ã®ã«ãã¯ã¢ãããã¼ãã«ãç¨æãã¦ããã¾ããï¼ãã¼ãã«ã®ä½¿ãæ¹ã®è©³ç´°ã¯å¾ã»ã©ï¼ ããã¦ãç»åã®å·¦ä¸ããã©ã¹ã¿ã¹ãã£ã³ãè¡ããç»ç´ ã®è²ãç½ã®ä½
MPS æ¦è¦ MPSã¯ããã¿ããªã§ããããã½ããã¦ã§ã¢ä½ããªããæè¡ã磨ãã¦å®ç¸¾ãä½ããï¼ããã³ã³ã»ããã«æ´»åãã¦ããã¾ããããã°ã©ãã³ã°ç人ã¨ãåå¿è ãã¿ããªã§ååãã¦ä½åä½ããè¡ããã§ãããã£ãä½åã¯å ±åå¶ä½ã®ãªã¼ãã³ã½ã¼ã¹ã»ã½ããã¦ã§ã¢ã¨ãã¦ãä¸ã®ä¸ã«åºãã¾ãã ç¾å¨ã¯ãä¸è¨ã®3ã¤ã®ããã¸ã§ã¯ããè¡ã£ã¦ãã¾ãã ã¨ã¸ã½ã³ã¾ãã¯ã©ãºããªã¼ãã¤ã¨ç»åå¦çãç¨ããï¼åï¼èªç«èµ°è¡ããããããã¥ã¢è»ãä½æ ã©ã¤ãé ä¿¡ã«é¢ããããã¸ã§ã¯ã MPS ã® HP 大æ¹ä¿® ä»åã®ä¼ç»ã¯æåã®2ã¤ã®ããã®åºç¤ç¥èãã¤ããåå¼·ä¼ã«ãªãã¾ãã åå¼·ä¼ã®å¾ã¯ã¿ããªã§ã飯ãé£ã¹ã«ããã¾ãï¼ ä»åã®äºå® Python ã§ç»åå¦çããã£ã¦ã¿ããï¼ï¼ç¬¬12åï¼ï¼ãã¬ã¼ã³ã¿ã¼éåï¼ ååã¯ãããã¾ã§ã«æ§ç¯ãã Scale space ããã¼ã¹ã« SIFT ç¹å¾´éãè¨è¿°ãããã¼ãã¤ã³ãåè£ã®æ½åºãè¡ãã¾ãããä»åããããã®
ä»åã¯David Loweã1999å¹´ã«ææ¡ããSIFTã«ã¤ãã¦è©±ãã¦ããã¾ããä»æ´ãSIFTãªãã¦è§£èª¬ãã¦ããããã«ãã£ã±ã解説ãããã ããã¨æãããããã¾ããããã¾ããããã¿ãªè¡ãã¾ããããããªããSIFTã«ã¯ç¹è¨±ãããã®ã§ä½¿ç¨æã«ã¯æ°ãã¤ãã¦ä¸ãããã¾ãããªã¼ãã³ã½ã¼ã¹ã®å®è£ ã«OpenCVã®ä»ãOpenSIFTãããã¾ããã¨ããã§ãæJAVAã§ä½ã£ãSIFTã®ããã°ã©ã ãè¦å½ããã¾ããã(T_T)ãããã¤ã¯ã©ãè¡ã£ã¦ãã¾ã£ãã®ãâ¦ã Scale-Invariant åå話ããããªã¹ã®ã³ã¼ãã¼æ¤åºã®ææ³ã«ã¯ãå転ä¸å¤æ§ãããã¾ãããç»åãå転ãã¦ãç¹å¾´ãå¤ãããªãããã¨ãããã®ã§ãããã§ããã¨ã ããã¢ãã£ã³å¤æä¸å¤ããã¹ã±ã¼ã«å¯¾å¿ããæ¹è¯ããããã¨ã ãè¨ãã¾ããããã«ãåºæ¬çã«ã¯ãã®ä»ã®å¤å½¢ã«å¯¾ãã¦ä¸å¤ãªç¹å¾´ã§ã¯ããã¾ããã ç¹ã«å°ãã®ããç»åã®å¤§ããã®éããå¸åã§ããªããã¨ã§ãããã¸
ãç®æ¬¡ã ï¼ï¼ï¼£è¨èªåºç¤ ï¼ï¼ï¼ï¼æ¬å½ã®åºç¤ ï¼ï¼ï¼ï¼é åã¨ãã¤ã³ã¿ ï¼ï¼ï¼ï¼æååæä½ã»ãã¡ã¤ã«æä½ ï¼ï¼ç»ååºç¤ ï¼ï¼ï¼ï¼ç»åãã©ã¼ããã ï¼ï¼ï¼ï¼ããã¹ãã¨ãã¤ã㪠ï¼ï¼ï¼ï¼é åã¨ãã¤ã³ã¿ ï¼ï¼ç»åå¦çåºç¤ ï¼ï¼ï¼ï¼ã¨ãã¸å¦ç ï¼ï¼ï¼ï¼èæ¯å·®åå¦ç ï¼ï¼ã°ã©ãæç»åºç¤ ï¼ï¼ï¼ï¼ï½ï½ï½ï½ï½ï½ï½ ï¼ï¼ï¼ï¼æãç·ã°ã©ã ï¼ï¼ï¼ï¼ãã¹ãã°ã©ã 表示 ï¼ï¼ã¢ã«ã´ãªãºã åºç¤ ï¼ï¼ï¼ï¼ï½ï¼å¹³åã¢ã«ã´ãªãºã ï¼ï¼ï¼ï¼ï¼¥ï¼ã¢ã«ã´ãªãºã ï¼ï¼ç»å表示åºç¤ ï¼ï¼ï¼ï¼ï¼¯ï½ï½ ï½ï¼§ï¼¬ ï¼ï¼ï¼ï¼ï¼¯ï½ï½ ï½ï¼§ï¼¬ã«ããäºæ¬¡å 表示 ï¼ï¼ï¼ï¼ï¼¯ï½ï½ ï½ï¼§ï¼¬ã«ããä¸æ¬¡å 表示 ã¯ããã« ããããç»åå¦çã»èªèã®ç 究ãå§ãããã¨ãã人ã対象ã¨ããå ¥éæ¸ãä½ã£ã¦ãã¾ãï¼å¯¾è±¡ã¯ç 究室ã«é å±ãããã°ããã®æ å ±ç³»å¤§å¦ã®ï¼å¹´çãæ³å®ãã¦ãã¾ããï¼èª°ãèªãã§ãåããããã«å¿ããã¦ããã¤ããã§ãï¼èªã¿é²ããªãã課é¡ã解ãã¦ãããã¡ã«ç»åã®åºç¤ç¥è
3.3. Scikit-image: ç»åå¦ç¶ èè : Emmanuelle Gouillart scikit-image ã¯ç»åå¦çã«ç¹åãã Python ç»åã©ã¤ãã©ãªã§ã NumPy é åãç»åãªãã¸ã§ã¯ãããã¤ãã£ãã«æ±ãã¾ãããã®ç« ã§ã¯ scikit-image ãå¤æ§ãªç»åå¦çã¿ã¹ã¯ã«ã©ãå©ç¨ãããã NumPy ã Scipy ãªã©ã®ä»ã® Python ã®ç§å¦æè¡ã¢ã¸ã¥ã¼ã«ã¨ã®é£æºã«ã¤ãã¦ãæ±ãã¾ãã åè åºæ¬çãªç»åæä½ããã¨ãã°ç»åã®åãæããåç´ãªãã£ã«ã¿ãªã³ã°ãªã©ãå¤ãã®åç´ãªæä½ã¯ NumPy ã SciPy ã§ãå®ç¾ã§ãã¾ã Numpy 㨠Scipy ãå©ç¨ããç»åã®æä½ã¨å¦ç ãåç §ãã¦ä¸ããã ãã®ç« ãèªãåã«åã®ç« ã®å 容ã«ã¤ãã¦æ £ãã¦ããå¿ è¦ãããã¾ãããã¹ã¯ãã©ãã«ã¨ãã£ãåºæ¬æä½ã¯æºåã¨ãã¦å¿ è¦ã§ãã
2.6. Numpy 㨠Scipy ãå©ç¨ããç»åã®æä½ã¨å¦ç¶ èè : Emmanuelle Gouillart, Gaël Varoquaux ãã®ç¯ã¯ãç§å¦æè¡è¨ç®ã³ã¢ã¢ã¸ã¥ã¼ã«ã§ãã Numpy ã Scipy ãå©ç¨ããç»åã«å¯¾ããåºæ¬çãªæä½ã¨å¦çã«ã¤ãã¦æ±ãã¾ãããã®ãã¥ã¼ããªã¢ã«ã§æ±ãããã¤ãã®æä½ã¯ç»åå¦ç以å¤ã®å¤æ¬¡å é åã§ãå½¹ã«çµã¤ã§ããããç¹ã« scipy.ndimage 㯠n-次å ã® NumPy é åãæä½ããé¢æ°ãæä¾ãã¾ãã
( ) ( ) (An Introduction to the Kalman Filter by Greg Welch and Gary Bishop) 1,2,, ,k-1,k,k+1,, ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )( ) (estimate) ( ) (1) (t=t0) (t>t0) (2) (t=t0) (t=t0) (3) (t=t0) (t<t0) ( ) (t=t0) (t=t0- t) (t0- t<t<t0) ( ) ( ) ( ) â â â â ( ) x k A k x k B k u k w k + + + + = = = = + + + + + + + + 1 (1) M q && x && + = δ0 0 (2) A k B k u l â â â â (2) z m â â â â ( ) ( ) (1) (2) w k
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}