Fusion Algorithm For Multi-focus Images Based On Saliency Region

Image is a kind of information carrier that contains rich data.For people,using images to express information is more direct.This way is more conform to people's habits.In recent years,technology of optical sensor has developed rapidly.According to different environment and different needs,we can select corresponding types of sensors to get images with specific information.For example,in the field of medicine,there are CT images that reflect human structure and MRI images that reflect human soft tissues.In the field of remote sensing image,there are high resolution panchromatic images,but the wave band of the image is single and the whole color is gray,and there are many multi spectral images which with many bands and rich in colors,but the resolution is low.In actual research,researchers need to combine the imaging results of multiple optical sensors for obtaining the characteristics of target objects.This method can get more information,but it needs to process multiple images.Even the basic work of image matching will also become more and more complicated due to the increase of the number of images.Until the end of the twentieth Century,image fusion technology appeared.The purpose of image fusion is to process image information from multiple channels to integrate into a single image.This paper focuses on multi-focus image fusion algorithm,which is an important branch of image fusion.The algorithm is mainly aimed at a group of images with distinct regions due to different focus when shooting the same scene.The appearance of this kind of image is mainly because there are many objects in the same picture at the same time,as the distance between the object and the sensor is different,the sensor can not focus all the objects at the same time.Each imaging only ensures that the focus area is clear,and the other areas become blurred.In order to get clear images of all objects in the area,we can only get multiple images or image sequences by focusing separately.The purpose of multi-focus image fusion is to use the designed algorithm to keep the clear region of each source image into the fused image as much as possible and get a clear multi focus fusion image.Firstly,this paper introduces and analyzes the concepts and research status of multi-focus image fusion algorithm and explains the principle of wavelet decomposition in the multi focus image fusion algorithm based on wavelet decomposition in detail.Aiming at the characteristics that the high frequency sub-band of the image contains more details after the wavelet decomposition.In this paper,a new concept and method of saliency area is proposed,which is used to measure the details of image blocks.Then,the new computing method of saliency is applied to the high frequency sub-band fusion of multi-focus image fusion.In the low frequency sub-band part,we use sparse representation to fuse.The experiment shows that the new fusion rules are equal to most contrast algorithms in visual effect,and are superior to most of the contrast algorithms in detail retention,and most of the evaluation indexes are optimal.Next,paper optimizes the new fusion rule.Aiming at the low frequency sub-band fusion process,it has large time cost and poor interpretability.The idea of Poisson image processing is introduced,and the fusion rule of low frequency part is redesigned under the guidance of low frequency sub-band gradient information.The same image group is fused by using the improved fusion rules.The experiment proves that the improved fusion algorithm can achieve the purpose of multi focus image fusion.The visual effect is very close to the other algorithms.Compared with the improved algorithm,all the indexes have been improved,and there are obvious improvements in the two indexes based on the gradient calculation.It is proved that the use of Poisson image processing is effective.