CNN-LSTM Learning Approach-Based Complexity Reduction for High-Efficiency Video Coding Standard

Table 1

Simulation results of the proposed scheme versus original HEVC.

Class	Sequence	CNN-LSTM versus Original HM
		BD rate (%)	BD-PSNR (dB)	ΔT (%)
A (2560 × 1600)	PeopleOnStreet	1.70	−0.017	−48.88
	Traffic	1.53	−0.059	−66.38
Average class A		1.61	−0.038	−57.63
B (1920 × 1080)	Kimono	1.65	−0.052	−47.77
	ParkScene	2.79	−0.081	−70.82
	Cactus	1.73	−0.033	−53.85
	BQTerrace	1.75	−0.030	−65.62
	BasketballDrive	2.02	−0.045	−52.77
Average class B		1.98	−0.048	−58.16
C (832 × 480)	BasketballDrill	1.67	−0.061	−48.23
	BQMalls	1.38	−0.090	−48.07
	PartyScene	0.96	−0.038	−59.30
	RaceHorses	1.47	−0.055	−54.32
Average class C		1.37	−0.061	−52.48
D (416 × 240)	BasketballPass	1.26	−0.056	−56.67
	BQSquare	1.27	−0.046	−60.79
	BlowingBubbles	0.97	−0.034	−50.14
	RaceHorses	1.60	−0.018	−50.33
Average class D		1.27	−0.038	−54.48
E (1280 × 720)	FourPeople	2.71	−0.071	−72.42
	Johnny	2.46	−0.083	−73.59
	KristenAndSara	2.93	−0.094	−74.91
Average class E		2.7	−0.082	−74.64
Average		1.78	−0.053	−58.60