CAJ | 학술논문

针对锥束CT序列切片图像的增强问题,提出一种基于反拉普拉斯锐化掩模的CT图像自适应增强方法.首先计算图像各像素点的局部方差,并通过设定两个梯度阈值将图像按所含边缘细节的多少划分成3个区域,然后对不同区域赋以与局部方差关联的高频调节因子,以及与方向梯度关联的拉普拉斯掩模,从而实现图像的自适应各向异性增强.实验结果表明,该方法处理后的图像峰值信噪比较大,均方误差较小,细节更加清晰,对锥束CT序列切片图像增强有较好的适用性.
침대추속CT서렬절편도상적증강문제,제출일충기우반랍보랍사예화엄모적CT도상자괄응증강방법.수선계산도상각상소점적국부방차,병통과설정량개제도역치장도상안소함변연세절적다소화분성3개구역,연후대불동구역부이여국부방차관련적고빈조절인자,이급여방향제도관련적랍보랍사엄모,종이실현도상적자괄응각향이성증강.실험결과표명,해방법처리후적도상봉치신조비교대,균방오차교소,세절경가청석,대추속CT서렬절편도상증강유교호적괄용성.
An Unsharp Masking based enhance method using Laplacian for sequence of CT images is proposed. In this method, local variance of each pixel is computed, according to which and the two gradient thresholds the image is divided into three regions. Different region is given different enhance coefficient, which is determined by it’s local variance, and a Laplace template which is determined by corresponding gradient of current pixel. Thus, the adaptive and anisotropic enhancement of image is achieved. The results show, the enhanced images using proposed method have higher SNR, lower MSE and clearer details comparing which the traditional ones. The proposed method performs well in the enhancement of sequence of cone-beam CT images.