Subcellular Biological Particles Detection Of Fluorescence Microscopy Based On Compressed Sensing

In recent years, with the development of biomedical science, fluorescent microscopy images are routinely used in biology and molecular medicine to detect mitochondria in live cells. Detection of mitochondria in fluorescent microscopy images is one of the most important methods in the studies concerning apoptosis and the nature of life phenomena in the area of biomedical image processing.However, limited by hardware conditions of fluorescence microscopy and structure of live cells, biological microscopy sequence also contains mitochondria which are needed to detect and shadow of cytoplasm which can produce complex background structure. So the signal-to-noise ratio of this king of images is low. At the same time, the target particle appears as attached to the neuron axons of same size, and have no obvious boundaries. Therefore, such microscopy images which are lack of reliable recognition characteristics can no meet the traditional particle detection algorithm. How to detect particles in this kind of microscopy images becomes the difficult and hot research in the field of digital image.In this paper, we use wide field microscopy video sequence which is as the research object to put forward a new particle identification algorithm which is based on matrix decomposition method. This method can separate mitochondria from cytoplasm shadow to achieve the purposes of image enhancement and noise reduction. The experimental results show that this algorithm has realized the accurate detection particles in images.First of all, compressive sensing which has been widely used in the field of image processing. The theory breaks the traditional Nyquist-Shannon which is on the premise of reconstructing original signal accurately and reduces the sampling ratio greatly. Then, Matrix factorization theory which is based on the theory of compressed sensing is proposed. This theory can decompose a high dimensional signal into a low-rank component and a sparse component. This method is applied to deal with drosophila neurons microscopic image. The target mitochondria which can be as sparse component are separated from background noise which can be as low-rank component. At last, we do a simple introduction of basic steps of detection particles and different detection algorithms. IUWT particle detection algorithm and spot enhancing filter algorithm to identify images which are used different pretreatment method. The experimental results show that when we use matrix decomposition method to process images, the detection rate and detection accuracy rate improve greatly which is far higher than other algorithms. Therefore, the algorithm which is based on matrix factorization theory provides a efficient and accurate tool to study fluorescent microscopy images.