摘要
为了有效降低传统感兴趣区域(ROI)编码算法的复杂度,提出一种基于码率预分配的感兴趣区域编码算法。算法优先对ROI进行T1编码,并截取控制ROI编码输出码率大小,实现有损到无损的ROI编码。使用基于熵估计的码率预分配方法,确定背景区域(BG)所属码块的码率,并据此控制BG码块的T1编码深度,简化编码过程。实验结果表明,算法与比特平面提升和率失真斜率提升的ROI编码算法相比,大幅度减少计算量和存储量,适合硬件实现。
In order to decrease the complexity of the traditional algorithm for ROI coding, a new algorithm based on rate pre-allocation is proposed. The code block of ROI is primarily encoded, and its coding stream length is controlled to implement from lossy to lossless ROI coding. Certain bytes are allocated to each block of background regions（BG） according to their entropy value. In the light of the Mlocated rate, T1 coding process on the block of BG can be real-time terminated. As a result, the whole coding process is simplified. The experi- mental results show that the proposed scheme can efficiently reduce the cost of computation and memory usage compared with the bit-plane lifting and rate-distortion slope lifting ROI coding methods, which is suitable for implementation in hardware.
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2009年第4期745-749,共5页
Systems Engineering and Electronics
基金
国家自然科学基金资助课题(60532060
60802076)
关键词
图像压缩
感兴趣区域
码率分配
熵估计
image compression
region of interest
rate allocation
entropy estimate
