Research On Multi-focus Image Fusion Methods

Image fusion makes multiple images to generate one image according to the fusion rules which is larger information, stronger ability to identify, less uncertainty image than a single image before fusion. Since the camera imaging system is limited by the depth of field, the target scene of a range is in the focus, thus we can only obtain one clear target image. The method of multi-focus image fusion can get a full clear image through extract the clear target of some images.This paper focuses on multi-focus image fusion method which based on RPCA(robust principal component analysis) domain via SML Feature Detection and multi-focus image fusion method via regional growth and focus detection:(1) According to the characteristics of multi-focus image, the multi-focus image fusion method based RPCA domain and SML Feature Detection is proposed. Firstly, the multi-focus source image is decomposed into sparse image and low-rank image by RPCA. The focus point location of sparse image is located via SML. Secondly, The three directions consistency and morphological methods are put forward for optimize the focus detect result. Finally, we can get the fused image by focus decision figure. The proposed method not only improve the focus detection accuracy, but also optimize the focus detect result of the focus decision function.(2) According to the characteristics of multi-focus image that the target shape of image have irregularities. The multi-focus image fusion algorithm via region growing method and focus detection is proposed. Firstly, using wavelet transform to fusion image. Then, the contrast of bright and dark of initial fused image is enhanced by mathematical morphology method, and the enhanced image is segmented by region growing method. Secondly, the focus point of image is detected via gradient matrix of image and mean filtering method. Finally, the fusion image is obtained and detected by regional segmentation. The Results showed that the method simplifies the traditional focusing criterion and improve the edge of the image fusion results.