The invention relates to an energy-saving compressed sensing image coding method, which effectively solves the problem that the high coding energy consumption and codec energy consumption in the existing technology can not be taken into account. The method is: block segmentation of the input image, divide into several non overlapping image blocks, calculate the gradient value of each image block, and use the gradient values of each block. The measurement matrix is built and the CS measurement is carried out. The projection matrix is constructed by using the measurement matrix of each block, and the inner product of the projection matrix and the observation vector is implemented, and the image blocks are reconstructed and merged into the final reconstructed images. The invention implements adaptive CS measurement according to the gradient value of each image block, and the gradient value quantifies the sparsity of the image block, ensuring that the encoder can still capture most of the information of the image block under low energy consumption, and reconstructs the image blocks linearly by using the projection matrix, and reduces the decoding energy consumption while obtaining good reconstruction quality. Encoding and decoding of CIF format video sequences, encoding energy consumption is low, decoding energy consumption is low.
【技术实现步骤摘要】
一种节能的压缩感知图像编码方法
本专利技术涉及图像编码
,特别是一种节能的压缩感知图像编码方法。
技术介绍
随着物联网的高速发展,越来越多的设备实现了互联。设备间的数据传输消耗了巨大能量,因此,节能的物联网受到了学术界和工业界众多关注。视觉传感器是物联网中的主要能耗源,现有工作设计了一些节能传感器,例如,CITRIC(P.Chen,P.Ahammad,C.Boyer,etal."CITRIC:Alow-bandwidthwirelesscameranetworkplatform,"ACM/IEEEInternationalConferenceonDistributedSmartCameras,pp.1-10,2008)、MicrelEye(A.Kerhet,M.Magno,F.Leonardi,A.Boni,andL.Benini."Alow-powerwirelessvideosensornodefordistributedobjectdetection,"JournalofReal-TimeImageProcessing,vol.2,no.4,pp.331-342,2007)等。这些节能视觉传感器尺寸很小,仅有一美分硬币大小,但尺寸限制了计算能力,当实施计算量较高的图像编码算法,例如,JPEG(M.Pennebaker."JPEG:StillImageDataCompressionStandard,"SpringerScience&BusinessMedia,1992)、H.264/AVC(G.J.Sullivan,P.N.Topiwal ...
【技术保护点】
1.一种节能的压缩感知图像编码方法,其特征在于:对输入图像进行块分割,划分为若干互不重叠的图像块;计算各图像块的梯度值;利用各块的梯度值构造测量矩阵,实施分块CS测量;利用各块的测量矩阵,构造投影矩阵;实施投影矩阵与观测向量的内积运算,重建各图像块,并合并成最终的重建图像。
【技术特征摘要】
1.一种节能的压缩感知图像编码方法,其特征在于:对输入图像进行块分割,划分为若干互不重叠的图像块;计算各图像块的梯度值;利用各块的梯度值构造测量矩阵,实施分块CS测量;利用各块的测量矩阵,构造投影矩阵;实施投影矩阵与观测向量的内积运算,重建各图像块,并合并成最终的重建图像。2.根据权利要求1所述的节能的压缩感知图像编码方法,其特征在于,包括以下步骤:(1)、输入图像,对输入图像进行块分割,划分为若干互不重叠的图像块,方法是:利用CMOS传感器阵列,获得自然景观感知的图像x,由CMOS传感器阵列将自然景观感知的图像x输入到编码器中,对图像进行分块,划分为若干互不重叠的图像块,输入图像x,尺寸为N=Ir×Ic,对x作块分割,分块尺寸为B×B,并自上向下、自左向右排序图像块,各图像块按光栅扫描方式拉成列向量,其中第i个图像块表示为xi,i=1,2,…,n,n=N/B2;(2)、计算各图像块的梯度值,方法是:将图像块xi的四邻域块表示为xi○j,j=1,2,3,4,计算图像块xi与任一邻域块xi○j的能量误差,计算公式是:式中||·||2为l2范数,其定义如下:其中v是输入向量,vk为向量分量,L为向量长度。计算图像块xi与其四邻域块间能量误差的最大值Gi:Gi=max{Ei,1,Ei,2,Ei,3,Ei,4}式(3)式中max{·}表示取输入元素中最大值,利用Gi,计算xi的梯度值如下:(3)、利用各块的梯度...
【专利技术属性】
技术研发人员:李然,段晓萌,沈克琳,
申请(专利权)人:信阳师范学院,
类型:发明
国别省市:河南,41
还没有人留言评论。发表了对其他浏览者有用的留言会获得科技券。