Image quality optimization towards lidar registration based on iterative termination

Abstract

Image quality optimization is a key technique in image processing, whose goal is to improve image quality by image enhancement or image format transform. This paper aims at optimizing image acquisition using Lidar registration, which can cope with disadvantages of conventional algorithms such as low-resolution. Specifically, we propose an iterative termination optimization strategy based on image quality perception features and local mean estimation. First, fuzzy images with different types and degrees of distortion are incorporated to form a representative natural image set, and feature maps of fuzzy images are extracted by the natural scene statistical method in the spatial domain. Noticeably, the proposed algorithm which performs iterative deblurring operation records the optimal iteration point based on recording the quality value FSIM of the restored image, and calibrates the corresponding feature vector in the sample library with the optimal iteration point (step number). Afterwards, we leverage LME method to implement an estimate of the number of iteration steps. Based on these two steps, the estimation of the initial iterative monitoring point is completed, so that the subsequent adaptive iterative termination work is more purposeful to monitor the defuzzification metric. The optimization operation can be completed faster effectively.