【摘要】 提出一种基于MAS小波变换多尺度相关的红外小目标检测方法。该方法通过二进MAS小波对图像进行多尺度分析;利用边缘和噪声具有不同的Lipschitz指数造成它们的小波变换模在不同尺度下的不同传播特性,根据小波变换模尺度相关性区分边缘和噪声;利用小目标与背景和云层边缘具有不同的奇异性,在相同尺度下小波变换模不同的特性加以区分,得到最终的检测结果。实验结果表明,该方法能够有效地进行红外小目标检测。更多还原

【Abstract】 Based on MAS (Modulus Angle Separated) wavelet transform, a new algorithm of infrared small target detection was presented. The discrete dyadic MAS wavelet transform was employed to perform the multi-scale representation of image and multi-resolution analysis. Since noise and edges had different characterizations of Lipschitz exponents, they had different propagation characteristic in different scales. Normalization correlation modulus were calculated and compared with the modulus of wavelet transform, and then the edges and the noise were separated. On the other hand, the edges of small target, background and clouds had different singularity of the same scale. Adaptive thresholds were used to distinguish small target from the edges of background and clouds. The algorithm could both remove the effect of noise and weaken the edges of background and clouds. Experimental results show that the algorithm works efficiently in infrared small target detection.更多还原