Module 1: Digital Video Signal Processing Lecture 5: Color coordinates and chromonance subsampling. The Lecture Contains:

Transcription

1 The Lecture Contains: ITU-R BT.601 Digital Video Standard Chrominance (Chroma) Subsampling Video Quality Measures file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_1.htm[12/30/2015 4:10:11 PM]

2 ITU-R BT.601 Digital Video Standard This standard is developed to standardize the digital formats used to represent different analog TV video signals with a quality equivalent to broadcast TV. Spatial Resolution of BT. 601 Signal: The standard specifies digital video formats for both 4:3 and 16:9 IARs. We discuss only the version with aspect ratio 4:3 (Standard formerly known as CCIR-601 format). To convert a raster scan to a digital video signal, one need only sample the ID raster waveform. If total number of samples/line = The equivalent sampling rate samples / sec. In BT.601 standard, the sampling rate is chosen to satisfy two constraints, i. The horizontal sampling resolution should match vertical sampling resolution closely ie and ii. The same sampling rate should be used for PAL, NTSC / SECAM systems, and it should be a multiple of respective line rates in these systems. The first criterion calls for or This leads to MHz for NTSC MHz for PAL/SECAM The number closest to both above numbers and yet satisfies second criterion is The number of pixels per line are for PAL/SECAM. for NTSC and file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_2.htm[12/30/2015 4:10:11 PM]

3 These two formats are known as 525/60 and 625/50 signals reply. (Figure 8) The number of active lines per frame are 480 (21 to 263 and 283 to 520) and 576 (23 to 310 and 336 to 623) in 525 & 625 line systems respectively. But the number of active pixels/line is same & both equal to pixels. The rest are samples obtained during horizontal & vertical retraces, which fall in non active area 625/50 fields/sec. With BT.601 signal, the pixel width to height ratio is not 1, that is, the physical area associated with a pixel is not a square. Specifically, =8/9 for 525/60 and 16/15 for 625/50 signals. To display a BT.601 signal, the display device must have a proper PAR; otherwise the image will be distorted. For eq. when displayed on a computer screen with a PAR of 1, a 525/60 signal will appear stretched horizontally, whereas a 625/50 signal will appear stretched vertically. Ideally, are should resample the original signal so that (such that PAR = 1) For example, the 525/60 and 625/50 signals should be re-sampled to have 640 and 768 active pixels/line respectively. file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_3.htm[12/30/2015 4:10:12 PM]

4 Color coordinates and Chrominance subsampling BT.601 recommendations also define a digital color coordinate, known as YCbCr. The Y, Cb, Cr components are scaled and shifted versions of the analog Y, U and V components. The scaling + shifting are introduced so that resulting components take on values in the range (0,255). The spatial sampling rates for the chrominance components, Cb and Cr is usually only half of the sampling rate used for luminance Y i.e. [ information content of the chrominance components is considerably less than Y information] This leads to half the number of pixels in each line, but same number of lines/frame. This is known as 4:2:2 formats, implying there are two Cb and two Cr samples for every 4 Y samples. Chrominance (Chroma) Subsampling The HVS has poor response to chrominance spatial detail compared to its response to luminance spatial detail. The property can be exploited to reduce bandwidth requirements by subsampling the chroma components. The most commonly used chroma subsampling patterns are the following: 1. 4:2:2 subsampling: The chroma components are subsampled by a factor of two horizontally. This leads to half the number of pixels in each line but same number of lines/frame. This gives a reduction of 33% in overall data rate. (Figure 9) file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_4.htm[12/30/2015 4:10:12 PM]

5 2. In 4:1:1 subsampling: The chroma components are subsampled by a factor of 4 horizontally along each line. This gives a reduction of 50% in overall data rate. This subsampling method yields very asymmetric resolution in horizontal & vertical directions. (Figure 10) In 4:2:0 subsampling: The chroma components are subsampled by factor of 2 both horizontally and vertically giving a reduction of 50% in the overall raw data rate. (Figure 11) file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_5.htm[12/30/2015 4:10:12 PM]

6 Horizontally subsampled chroma samples can be either midway between luma samples or consisted with odd numbered luma samples. (Figure 12) To further reduce the required data rate, BT.601 also defined the 4:1:1 format in which the chrominance components are sub sampled by a factor of four along each line. This sampling method yields very asymmetric resolutions in horizontal and vertical directions. Another sampling format has therefore been developed called the 4:2:0. In this format, there is one Cb and one Cr for every 4 Y samples. For applications requiring high resolutions, the 4:4:4 format is defined. This format samples the chrominance components in exactly the same resolution as Y. A summary of different digital video formats with applications is given in Table 1 below. (Figure 13) file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_6.htm[12/30/2015 4:10:12 PM]

7 The raw data rate of a BT.601 signal depends on the color subsampling format.fig 7 shows the different color subsampling formats. We calculate the data rates for different formats in the following section. Consider 4:2:2 format. It has 2Cb, 2Cr samples for every four Y samples. Each sample rep by 8 bits. Equivalent bit rate for each Y sample is Raw data rate is Raw data rate corresponding to active area is For 4:2:0 format Equivalent bit rate for each Y sample = 12 bits. Raw data rate is 162 mbps with 124 mbps for active area; For 4:4:4 format:, Raw data rate is 324 mbps, and raw active data rate is 249 mbps. Table 1: Digital Video Formats (Figure 14) file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_7.htm[12/30/2015 4:10:13 PM]

8 Video Quality Measures Objective Criterion that can measure the difference between the original and processed signal. 1. Mean squared Error (MSE) between two video sequences and is defined as: N: total number of pixels in either sequence. For color video, the MSE is computed separately for each color component. 2. Peak signal to Noise ratio (PSNR) in db (used as quality measure in video coding) is defined as: Where is the peak intensity value of video signal usually taken as 255. PSNR is more commonly used than MSE. To compute the PSNR between two sequences, compute the MSE between corresponding frames, average the resulting MSE values over all frames, and finally convert MSE to PSNR. 3. Mean absolute Difference (MAD) is defined as:. file:///d /...rse%20(ganesh%20rana)/my%20course_ganesh%20rana/prof.%20sumana%20gupta/final%20dvsp/lecture5/5_8.htm[12/30/2015 4:10:13 PM]