Graphics Reference
In-Depth Information
We have implemented the video rendering in the virtual reality platform including
windows platform and android platform. Please note that not all the real video scene
is easy to record on the alignment angle position. Two ways can be selected to render
in the virtual reality system. As shown in figure 3, the real scene indoor is displayed
in a dialog by clicking the camera icon. For the real scene that is aligned in the accu-
rate angle position will be projected in the virtual reality system, which is more suita-
ble for outdoor scenes. Figure 4 has showed the live video projected in the system.
The scene object in the projection range shows the real situation of the current
position. We can also see the object out of the projection rangein the virtual reality
system.
5
Experimental Evalution
In this section we present a performance of hit rate experiment and load balancing test
in the current distributed architecture. The Experimental environment of the system
including the followings: a dell PowerEdgeR710 server (Intel Xeon 2.40GHz
16GB), as a streaming media server; a data server (Intel Core i5 3.10GHz
16GB);
three distributed server (Intel Core 2 Duo 2.80GHz
8GB); a test server (Intel Core 2
Duo 2.93GHz
4GB); a Samsung Galaxy i9300 mobile client and the HTC Onex
mobile client.
5.1
Hit Rate Experiment
The 40 GB disk space in the content disputation server is allocated to the cache video
data for experiment. 400 requests with more than 200 different requests from 10 users
are selected to make a simulation experiment. The size of the file is between 200MB
and 1G. The Cache is full and the contents are the same for each experiment. Hit rate
is counted with the function policy designed, LRU policy, LFU policy and size poli-
cy. As shown in figure 5, the weight function design has a better performance because
its hit rate is higher than others.
Fig. 5.
Hit rate in four experiments
Search WWH ::
Custom Search