1. Digital Video Solutions to Midterm Exam 2004 Edited by Yu-Kuang Tu Confirmed by Prof. Jar-Ferr Yang LAB: 92923 R, TEL: ext. 621 E-mail: specta@video5.ee.ncku.edu.twspecta@video5.ee.ncku.edu.tw Page of MPL: http://mediawww.ee.ncku.edu.twhttp://mediawww.ee.ncku.edu.tw

2. 2.1. (a) (b) (c) (d) 81 horizontal lines per frame, 2:1 interlace scanning format, aspect ratio of 4:3, 20 frames per second. vertical Horizontal …… 81 個

3. 2.2. (a) (b) (c) (d) (e) (f) 水平掃描電壓波幅不足 In Fig. g, H bias ckt has problem 垂直掃描電壓波幅不足 In Fig. g, V bias ckt has problem Clock 不同步 In Fig. g, V bias ckt has problem Video Amplifier (AGC) 有問題 In Fig. h, the video amp (upper one) has problem with improper DC offset. Blanking level 不正常 In Fig. h, the video amp (upper one) has problem Color decoder 有問題 In Fig. h, the video amp (lower one) or 2~4.2M BPF has problem normal abnormal vertical Incorrect time

4. 2.3. (a) (b) (c) A 1 A 1 0.766 A 1 A 0 0.11 A 0 A 1 0.11 A 0 A 0 0.0156 0.1256 0.2356 0.00.1251.0 0.1250.2341.0 0.2340.331.0 0.330.4141.0 A0A0 A0A0 A0A0 A0A0 A1A1 A1A1 A1A1 A1A1 0.33<W<0.414 W=3/8  011 Two-event (two examples) Huffman code: A 1 A 1 0 or 1 A 1 A 0 11 or00 A 0 A 1 100 or011 A 0 A 0 101 or010 1 0 1 0 0 1

5. 2.3. (d) W=0.375 A 1 A 1 A 1 0.00.1251.0 0.1250.2341.0 0.2340.331.0 A0A0 A0A0 A0A0 A1A1 A1A1 A1A1

6. 2.4. H0(z)H0(z) H1(z)H1(z) ↓2 G0(z)G0(z) G1(z)G1(z) ↑2 x[n]x[n]x[n]x[n] y[2n] y[2n+1]

7. 2.4. n -4-3-20123456789 x[n]x[n] 321234543 y 0 [2n] 50301030503010 y 1 [2n+1] 00000000000000 x0[n]x0[n] 12345 x1[n]x1[n] 00000 x’[n] 12345 The last row is the reconstructed synthesis data h0h0 h0h0 h0h0 h1h1 h1h1 g0g0 g0g0 g0g0 g1g1 g1g1 g1g1 g1g1 g1g1 g0g0 g0g0 -2 1 0 1 2

8. 2.7. 42 23 -26 19-12 14 1012 310 7-4 00 12 Initialization: LIP: { (0,0)  42, (0,1)  23, (1,0)  -26, (1,1)  12 } LIS: { (0,1)D, (1,0)D, (1,1)D } LSP: {} Significant Pass: 10,0,0,0 0,0,0 LIP: { (0,0)  42, (0,1)  23, (1,0)  -26, (1,1)  12 } LIS: { (0,1)D, (1,0)D, (1,1)D } LSP: { (0,0)  42 } Refinement Pass: 48 0 0 00 00 00 00 00 00 0 LIP LIS

9. 2.7. Significant Pass: 10,11,0 0,1;10.0.0.0,0 LIP: { (0,1)  23, (1,0)  -26, (1,1)  12, (2,1)  -12, (3,0)  14, (3,1)  14 } LIS: { (0,1)D, (1,0)D, (1,1)D } LSP: { (0,0)  26, (0,1)  23, (1,0)  -26, (2,0)  19 } Refinement Pass: 0 * 23 -26 19-12 14 1012 310 7-4 00 12 40 24 -24 240 00 00 00 00 00 0 LIS LIP

10. SPIHT Significant Pass: 10,11,10…. LIP: {(1,1)  12, (2,1)  -12, (3,0)  14, (3,1)  14 } LIS: { (0,1)D, (1,1)D } LSP: { (0,0)  26, (0,1)  23, (1,0)  -26, (2,0)  19 } Refinement Pass: … * * * *-12 14 1012 310 7-4 00 12 25bits 40 24 -24 24 (-)12 00 00 00 00 00 12

11. 3.1. 3.2. 3.3. 3.4. 3.5. F, …remove temporal redundancy…, …interframe coding… F, …spatial-domain data…., …a spatial redundance removal… F, RLC is a data compression technique. F, However, there still exist some ways depending on data property to find an alternative scanning order to improve the coding efficiency actually F, With the same rate, the hierarchical mode needs more computation and obtains less compression ratio.