View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000028 | JVT JM H.264/AVC reference software | decoder | public | 2006-12-25 09:49 | 2007-01-03 12:38 |
| Reporter | Zhou Ya | Assigned To | |||
| Priority | normal | Severity | trivial | Reproducibility | always |
| Status | closed | Resolution | no change required | ||
| Product Version | JM 10.0 | ||||
| Summary | 0000028: Caculate wrong PSNR when Frame Skip is used | ||||
| Description | steps to reproduce the error: 1)Set Encoder.cfg->Encoder Control->FrameSkip =1 or any other nonzero value. 2)Encoder a video sequence. 2)use JM10.0 Decoder to decode the compressed video file "test.264" with the default configuration file "decoder.cfg". The PSNR(Y) caculated by JM10.0 will about 20dB or even less,which is unreasonable.I've read the function find_snr in image.c and I think maybe the reason is the wrong position sought by lseek (p_ref, framesize_in_bytes * frame_no, SEEK_SET) because frame_no is not the real number of frame in test_rec.yuv but the number in the *.yuv uncompressed file. | ||||
| Tags | No tags attached. | ||||
|
|
The reference picture used for PSNR calculation can be selected in the decoder configuration file: 2 ........Reference POC gap (2: IPP (Default), 4: IbP / IpP) 2 ........POC gap (2: IPP /IbP/IpP (Default), 4: IPP with frame skip = 1 etc.) The decoder basically has to "guess" which picture of the original file was used while encoding. These decoder parameters should allow proper calculation for most settings of the JM encoder (except some IDR cases). I fix for the missing cases is already being developed. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2006-12-25 09:49 | Zhou Ya | New Issue | |
| 2007-01-03 12:38 | Karsten Suehring | Status | new => closed |
| 2007-01-03 12:38 | Karsten Suehring | Note Added: 0000041 | |
| 2007-01-03 12:38 | Karsten Suehring | Resolution | open => no change required |
