Đang chuẩn bị liên kết để tải về tài liệu:
O’Reilly Learning OpenCV phần 7

Thay vì âm mưu một chuỗi các giá trị cho mỗi khung hình (như chúng tôi đã làm trong hình đó), chúng tôi có thể đại diện cho các biến thể của mỗi điểm ảnh trong suốt đoạn video về một sự khác biệt trung bình và trung bình (Hình 9-2). | w w w w u IM V I.M IM d 2x1 IM p Apertu re p roblem Figure 10-8. Aperture problem through the aperture window upper row we see an edge moving to the right but cannot detect the downward part of the motion lower row We now have an overconstrained system for which we can solve provided it contains more than just an edge in that 5-by-5 window. To solve for this system we set up a least-squares minimization of the equation whereby min Ad -b is solved in standard form as AyA d gb y ucw 2x1 hi i im this relation we obtain our u and V motion componentt- writing this cut inmure detti i yields LVx 1M 1 . _ y ii yii V X y y yJL . Eul xu The solution to this equation is then u . T . _1 . T . . 328 I Chapter 10 Tracking and Motion When can this be solved when ATA is invertible. And ATA is invertible when it has full rank 2 which occurs when it has two large eigenvectors. This will happen in image regions that include texture running in at least two directions. In this case ATA will have the best properties then when the tracking window is centered over a corner region in an image. This ties US back to our earlier discussion of the Harris corner detccto r. In fact those cornens ww good features tot ack see our previous remarks conoerning cuGcudFccSunesOuSracrOC tor preciseSy Iha reaconthat lATht hud two Inge cige nveeFonstUere Ws hsse rhorily how e 11 e h-s aompstFtson is done forus by the aCrlcOpticalFSouLKf tunrFion. cFie seader who underrtends the impStsrtionr oF sus asesiming stnFO acdcoherrnt mo-oisnsw now b kdolhcFt d by OhrfocS shah fos most vtdeo romerat sunning ht Fh Ha lergr n ny nonooforent motions are commonjalr co. In t ct .a cactKonudr o ira 1 flowOy itsela doao hoc work aray well foa nunctly thisrecsup we want a h rgswm tlnw tocateh large motiooSt but a 1 argo window too often breaks the coherent motion assumption Th c ĩacc i vo t t hio proutom wtcan track fiottkvns 1 argsr epottnr rcatco osínu aa émeye pyramid sod then FeUar sttr i nUiai moticn v Hg assumptinns