A method of automatic ISAR Image quality evaluation

It is of great significance to select high quality ISAR imaging results from massive data processing results to effectively obtain target information for target recognition. In this paper, an automatic quality evaluation method for ISAR images is proposed, which can automatically filtering and excluding low-quality images. Firstly, the quality evaluation standard image database was established by expert evaluation and objective model evaluation. Then, the features of ISAR images are extracted by convolutional neural network. Finally, SVM classifier is used to evaluate the image quality. Experimental results show that the proposed method is more reasonable, accurate and efficient than the objective image quality evaluation method without reference.