Edge Detection Algorithm Research For Biological Images

In the wake of the interplay between information discipline and modern life science, digital image processing technology is widely applied to biological quantitative analysis in recent ten years. Edge detection is the necessary pre-technology for biological quantitative analysis. In particular, the quantitative analysis of the cell slice color microscopic images and ultra-microscopic images is a difficult aspect of biological quantitative analysis. Therefore, specifically for the edge detection issues of cell slice color microscopic image, this paper carried out the research work as follows:1. At first described and analyzed the basic characteristics of the simplified PCNN model, and then the entropy sequence characteristic of PCNN is applied to fingerprint feature extraction. Experiment results indicate that as a fingerprint feature, the entropy sequence has a good identity but also unsatisfactory exclusive nature. In this paper, the size of internal matrix has been changed and experiment results show changing the size of internal matrix could improve the exclusive property.2. Proposed the improved pulse coupled neural networks and extended the PCNN application and made it directly deal with cell slice color microscopic image. The improved vector model is more suitable for color image processing, because each pixel of color image can be seen as a vector. Based on analyzing the research status and difficulties of the microscopic cell image edge detection, this paper put forward a PCNN method to get a clear continuous cell edge form the microscopic cell image with a complicated background and gave out the graphical user interface for the proposed PCNN method. The algorithm could directly process color images and was far superior to traditional edge detection algorithm on weak edge detection. In particular dealing directly with color images is a bright spot of the work and is a pioneering application of PCNN. In addition, experimental results have shown that the algorithm inherited the strong anti-noise of PCNN. This paper also focused on how the important parameters of the edge detection algorithm impact the efficiency of the algorithm.3. On the basis of introducing the basic theory of quaternion and summarizing the basic idea for applying quaternion to color edge detection, this paper presented an edge detection algorithm for microscopic cell image with a complicated background. Experiments results demonstrated that the algorithm could obtain distinct continuous color edge of the cell images. Since the algorithm performed an integrated treatment of color images rather than process every channel separately, it has competitive advantage of retaining the association between channels as vector analysis method. This is a bright spot of the paper work.