Computer Processing Method Of Gel Electrophoresis Image

SSR(Simple Sequence Repeat)molecular masker technology has become the core means in the field of agronomy to identify the genetic relationship between different varieties of crops,the classification of genetic relationships,the evaluation of core germplasm resources,etc.,but in the actual research process,the experiment obtained gel electrophoresis images often have many influencing factors such as noise and smear between bands.Since the final analysis result depends on the band analysis result of the gel electrophoresis image band,the use of computer technology to clarify the gel electrophoresis image band and the automatic identification of the gel electrophoresis band has always been an important topic.In view of the above problems,we studied the method of remapping and strip recognition in gel electrophoresis.Specifically,a redrawing method of gel electrophoresis image is designed and implemented based on Open CV.First,the gel electrophoresis image is preprocessed by affine transformation,denoising,grayscale and two valued algorithm,and then image repainting is done on this basis.Deep learning framework is deep designed based on the area convolution neural network algorithm.A deep learning framework for detecting gel electrophoresis image strips is designed.Two abnormal strip correction algorithms are designed according to the characteristics of gel electrophoresis images.Get an improved gel electrophoresis image band detection algorithm based on Mask R-CNN.The improved algorithm is compared with Sobel operator.The experiments show that the recognition method proposed in this thesis obtains a higher F1 measure value.