Lenses matching of compound eye for target positioning

Abstract

Compound eye, as a new imaging method with multi-lens for a large field of view, could complete target positioning and detection fastly, especially at close range. Therefore it could be applicated in the fields of military and medical treatment and aviation with vast market potential and development prospect. Yet the compound eye imaging method designed use three layer construction of multiple lens array arranged in a curved surface and refractive lens and imaging sensor of CMOS. In order to simplify process structure and increase the imaging area of every sub-eye, the imaging area of every eye is coved with the whole CMOS. Therefore, for several imaging point of one target, the corresponding lens of every imaging point is unkonown, and thus to identify. So an algorithm was put forward. Firstly, according to the Regular Geometry relationship of several adjacent lenses, data organization of seven lenses with a main lens was built. Subsequently, by the data organization, when one target was caught by several unknown lenses, we search every combined type of the received lenses. And for every combined type, two lenses were selected to combine and were used to calculate one three-dimensional (3D) coordinate of the target. If the 3D coordinates are same to the some combine type of the lenses numbers, in theory, the lenses and the imaging points are matched. So according to error of the 3D coordinates is calculated by the different seven lenses numbers combines, the unknown lenses could be distinguished. The experimental results show that the presented algorithm is feasible and can complete matching task for imaging points and corresponding lenses.

Research on key technology of cooled infrared bionic compound eye camera based on small lens array

“…In terms of lens array camera design, in 2012, Brady et al described the AWARE-2 camera, which has a single aperture with a diameter of 16 mm and can capture 1 gigapixel images at three frames per minute, and AWARE-2 solves the problem by using a parallel microlens array that converts gigapixel images to megapixel images 20 ; In 2012, the Wang KY team proposed a new large field of view multi-lens imaging method using a three-layer structure of curvilinear multi-lens arrays and matched refractive lenses and an imaging sensor CMOS that establishes the matching relationship of adjacent lenses based on geometrical relationships to calculate the 3D coordinates of the target 21 ; In 2013, Afshari et al proposed a new bionic vision sensor, Panoptic. Suitable for real-time acquisition and prIocessing of 3D image sequences, the device consists of a hierarchical arrangement of about 100 CMOS imagers distributed on a hemispherical plane with a diameter of 13 cm, all of which have different focal points and different fields of view, allowing the application of specific signal processing for the acquisition of 3D information, such as omnidirectional stereo or depth estimation 22 ; In 2015, Cao et al proposed a design method for a spherical compound eye consisting of 37 imaging channels, where the device consists of multidimensional sub-magnitude imaging channels with different observation directions, and the corresponding imaging experiments were performed in combination with the split-rotation projection technique, which verified an observation field of view of up to 118 •23 ; In 2017, Shi CY et al demonstrated a three-part hemispheric compound eye, an optical relay system, and a commercial CMOS imaging sensor hemispheric compound eye camera (SCECam), which can achieve a large observation field of view of up to 122.4 • and can detect and locate the speed of fast-moving objects 24 .…”