Double-camera system and method for actively acquiring high-resolution image of interested target

Abstract

The invention discloses a double-camera system and method for actively acquiring a high-resolution image of an interested target. The double-camera system comprises an implicit attention camera, an interested target finding module, a multi-target attention scheduling module, a position estimating module, a field-of-view coupling module, a deviation control module and an explicit attention camera. The method comprises the following steps: monitoring the global scene on real time through the implicit attention camera in order to find out the interested target from a dynamic scene; calculating the attention priority of each target; focusing the attention of each target and performing real-time scheduling based on the priority; estimating the position of the selected target; compensating the system time lag; performing field-of-view coupling calculation of the implicit attention camera and the explicit attention area, and converting the spatial property information of the target into fixation state parameters; finally controlling the explicit attention camera to acquire the high-resolution clear image of the target based on the fixation state parameters. With the adoption of the double-camera system and method for actively acquiring the high-resolution image of the interest target, the problem of low resolution of a mini target object in the large scene monitoring can be solved.

Description

technical field [0001] The invention belongs to the technical field of computer vision application and intelligent monitoring, and in particular relates to a dual-camera system and method for actively acquiring target high-resolution images. Background technique [0002] Traditional video surveillance systems passively monitor the scene. According to the camera imaging model, the larger the three-dimensional scene corresponding to the scene monitoring image, the smaller the area occupied by the monitoring target object in the image plane; in addition, the CCD unit size resolution is limited, which eventually leads to the small target object in the large scene monitoring video due to the resolution Too low and difficult to identify, making surveillance video images lose their value as identification clues and evidence. [0003] To solve the above problems, there are two ways to improve the image resolution of monitoring target objects: one is to use super-resolution and othe...

AI Technical Summary

Problems solved by technology

[0005] In view of the above-mentioned problem of low resolution of tiny target objects in large scene monitoring, the present invention proposes a dual-camera system and method for actively acquiring high-resolution and clear images of objects of interest in the scene while monitoring the global scene in real time. The source of the acquisition performs selective perception of the target of interest, effectively improving the target resolution

Images