Paper Information

Journal:   AMIRKABIR   Winter 2004 , Volume 15 , Number 57-A; Page(s) 1 To 16.
 
Paper: 

NEW MATCHING CRITERIA FOR VIDEO COMPRESSION APPLICATION: APPROPRIATE FOR HARDWARE IMPLEMENTATION

 
 
Author(s):  POURREZA H.R., RAHMATI M., BEHAZIN F.
 
* 
 
Abstract: 

Estimation of motion is an important step in video compression applications. Block matching motion estimation technique is one of the approaches that are widely used for this purpose. In this technique, by incorporating a matching criterion and employing a search strategy, the translational motion is estimated on the image blocks. In this paper, we present three new criteria, which are suitable for hardware implementation. In two of these introduced criteria incorporated an existing criterion and modified it for a better performance. These criteria are based on bit-plane matching where four sub-sampled bit-planes were used. Basically, these two are based on a criterion which is presented for digital image stabilization application. In this paper, while this were modified criterion for video compression we investigated the cause of problems and suggested the possible solution. Our experimental results indicated that the two proposed criteria yield better performance than the original method, where in the first case the performance increased by %6 and for the other case the growth is %9. Our third criterion is the adaptive form of pixel difference classification (PDC). In our proposed method, two modifications were proposed on the basic PDC, in which for one of them we reduced computational burden and in the other the performance we increased. Our experiments indicate that our proposed adaptive approach can yield %99 of PDC performance when used the best threshold. only need about 0.0056 additional operations per pixel for obtaining an adaptive form of PDC. But the modification that is presented for reducing the computational burden can compensate these additional computations.

 
Keyword(s): DIGITAL VIDEO STABILIZER, MOTION DESTIMATION, MOTION COMPENSATION, MACHINE VISION, DIGITAL VIDEO IMAGE PROCESSING, VIDEO COMPRESSION
 
References: 
  • ندارد
 
  Persian Abstract Yearly Visit 64
 
Latest on Blog
Enter SID Blog