The Research Of Microscopic Image Sequences Coding Method

Biological microscope images are collected from slices of animal or plant tissue by digital microscope. In general, the microscopic image owns high-resolution features, which present the storage and transmission microscopic image a serious problem. Efficient compression of the images is a key technology to solve this problem.First, in the paper we use a variety of compression methods to encode the microscopic image sequence, and analyze the performance of various methods. Considering the micro-image characteristics, we provide novel multi-resolution compression arithmetic based on H.264 Standard. First encode the lower resolution image by H.264 encoder normally, then obtain the texture objects using the residual data between the original and the reconstructed image of the small resolution and code them finely. The experiment shows that this arithmetic can get a better compression ratio and reduce the complexity of the encoder while it maintains the same object quality.Secondly, taking into account the microscopic image of the global motion features, this paper presents several global motion estimation algorithms. Advantages and disadvantages of various algorithms are analyzed. Experiments show that the overall iterative multi-resolution motion estimation algorithm is the most stable way, but the computation is larger than the full-search method. By randomly selecting a fixed number of specific macro block to search has the minimum amount of computation. Experiments show that the iterative global motion estimation method saves much time compared to the method that used in the MPEG-4 standard.Finally, in order to solve the restrictions on microscopic image sequence of small field when view microscope images , splicing has been studied and put forward an adaptive mark selection algorithm, experiments show that the algorithm can adaptively choose a splice location to eliminate the noise impact. Using the global motion vectors to registration, filter the splice location by mean filter can well eliminate the patchwork of the visual impact.