No abstract

Abslruel-Neal-the Ininge-bawd rendering tcclinlques for the real rnvironmcnt arc proposcd. The tcchnlquer, pwered by AKl'oolklt, conaldcr an oliject's re!lcctlon, tmnnluccncy, shadow, slinde and color halancc under real Ilghtlng. Expcrinicntnl results are also shown.One of the issues in augn~entcd reality is highly realistic real timc rendering of virtual objects 111 thc red world. Por this purpose, wc have dcveloprd image-bared rendering techniqucs that arc suitable fur the rendering part of ARToolKit. The proposed techniques can render object iinagcs railistic cnoiigh to I n~n~n vision in (lie rral enviruniiient in real time. 11. IUAL-TML IMAOE-BAS~D TKllNIQl1F.S FOR E N V I R O N M~N T MAPPINGWe assuinc that a virtiial object is collectiun of iiiultiviewpoint images cif a real object with tlie background dcletcd. We have developed a C ' C~J~I~I X .system [3] for generating multi-viewpoint iiiiages fur real objects. The systcin takes photugapbs of an object from verious angles and ciui autoo~atically gciicratc alpha masks and texltues described in section 11-A within about 30 minutes. At the timc ofrendering, the image closcst to the uscr's viewpoint is selected and aupcrimposcd at the registered position by the AWoolkit. In the rest of this section, we describe rendering techniques tu match these objcct images to the cnvironment. A.Accounting for rcflectioiis and translucency lras proven to bc very imporhiit in tiying tu create reiilistic images of v~rtual objects that hiiiiiionize with the C I I V I~U I~I I~C~~. We have developed new image-based icndcring tccliniqiic called aBBB (Alptm Blcnrliiig ivrth Bliirred Bockgroioid). which represents pseudo-reflection and transparency in rcal time.The complexity of ;scuiatc caIci~IaI~u~i of retlection or translucency. such as lay tracing, is vciy high. and impussible for ordinary PCs tu process in real titnc. On the olhcr hand, it is known that liiinlans do not recognize d l of aspects (if images. Our psychophysical experiinents have rliown that moat of the reflections of glossy objects can be simiilatcd by ;illowing tlic siimc quantity of reflection components ;IS there Hepi.eserrtnliotr of Rcflrcliorl a~l d Tronsporemj (c) blurrcd b;ickground (d) synthetic image Fig 1. l h e aBUB Tecliniquc arc it1 tlie blurred backgmond to pass through the object [I], The aBBB is based on tlic rcsiilt of this observation.The aBBB uscs two images, tlie texture of the object and the corresponding alpha mask such as showi~ in Fig. 1 (a) and (b). The alpha nlask indicates the amount of reflection and transpirency in tlic gray level. In the aBBB, first the background image is blurrcd according to the pixel v~l i i c of the alplia mask by applying a low-pass filter (Fig. I(c)). The texture is tlien superiiiiposcd on new background by mentis of alpha blending (Fig, I(d)). B. Mulching f$Lig/if Soiirce DirectionThe iippenraiice of objects clianges greatly depcnding on the direction uf tlie light suurcc. The object image must be adjusted to match the light source directiun in the ciivironnic...

SUMMARYHardware acceleration is an essential technique for extracting and tracking moving objects in real time. It is desirable to design tracking algorithms such that they are applicable for parallel computations on hardware. Exclusive block matching methods are designed for hardware implementation, and they can realize detailed motion extraction as well as robust moving object tracking. In this study, we develop tracking hardware based on an exclusive block matching method on FPGA. This tracking hardware is based on a two-dimensional systolic array architecture, and can realize robust moving object extraction and tracking at more than 100 fps for QVGA images using the high parallelism of an exclusive block matching method, synchronous shift data transfer, and special circuits to accelerate searching the exclusive correspondence of blocks.

No abstract

No abstract