论文部分内容阅读
针对前视声纳图像清晰程度不同,局部区域模糊的特点,提出一种基于NSCT多尺度分析的前视声纳图像融合算法。首先,对源图像进行NSCT多尺度分解,得到一系列多尺度子带分解系数;然后,根据图像中清晰目标反射声波能量大、对比度高特点,构建前视声纳图像融合规则:分别对低频子带采用Gabor能量,高频子带计算局部对比度指导融合规则,并提出区域一致性校验准则抑制图像噪声,产生融合图像多尺度子带分解系数。最后,应用NSCT逆变换获得融合图像。声纳图像融合对比实验证明提出方法有效性。
Aiming at the different sharpness of foreground sonar images and the fuzzy features of local areas, a forward-looking sonar image fusion algorithm based on NSCT multi-scale analysis is proposed. First, the NSCT multiscale decomposition of the source image is carried out to obtain a series of multi-scale sub-band decomposition coefficients. Then, according to the characteristics of large target reflected acoustic energy and high contrast in the image, the frontal sonar image fusion rules are constructed: The band adopts Gabor energy and high frequency sub-band to calculate the fusion rule of local contrast, and proposes the regional consistency check criterion to suppress image noise and produce multi-scale sub-band decomposition coefficient of fusion image. Finally, the NSCT inverse transform is used to obtain the fused image. Sonar image fusion experiments show the effectiveness of the proposed method.