Lens calibration for obstacle detection
- PDF / 632,634 Bytes
- 6 Pages / 595.276 x 790.866 pts Page_size
- 76 Downloads / 277 Views
RESEARCH ARTICLE
Lens calibration for obstacle detection Y. V. Chavan1
•
P. Y. Chavan2 • A. Nayanit1 • V. S. Waydande1
Received: 27 April 2020 / Accepted: 13 July 2020 Ó The Optical Society of India 2020
Abstract Lens is an important component of any imaging and visual system, when light falls on any object, partly it is reflected, partly transmitted through the object. These systems are based on the light reflected from the object(s), which consists of spectrum of different wavelengths of the object. Light reflected from object, transmitted through the lens and refracted thereafter. All the imaging system is based on the refractive properties of the lens, analogous with tolerances and limitation such as aqueous humor in cornea, and vitreous humor in lens of Human Visual System, HVS. The image formation by the lens system can be effectively extended for automated vehicle application for obstacle detection. This paper emphasizes on lens modeling, considering internal, external parameters of lens and performance parameter along with the images registration on CMOS image sensor, simulation is presented for hardware. The CMOS image pixel sensor can be thought and extended as short term memory while making the decision in order to avoid obstacle within its time frames. Keywords Lens calibration Lens modeling CMOS digital image pixel sensor Stereo camera images
& Y. V. Chavan [email protected] P. Y. Chavan [email protected] A. Nayanit [email protected] 1
Government Polytechnic, Karad, Dist. Satara, MS 412125, India
2
InfinityX Pvt. Ltd, Pune, India
Introduction Human is always trying to mimic the natural system, where the automation is required based on the sensors and technology available. HVS is always being the hot topic for automation, as human processes most of the happenings around are through eyes only. The same is true for artificial system comprising of vision for application. Modeling and simulation of lenses for Retinal system [1] have discussed many aspects of HVS. Basically, it is the limitations of material science in wider-sense; the technology can not exactly mimic the natural system [lenses for vision]. One of the major limitations is the refractive index as the requirement for HVS is shown in Table 1, while the water has refractive index of 1, the rest all have widely varying as compared to the requirement of around 1.33 in human vision. This paper discusses the issues of 2-D and 3-D lens modeling and then extension of this for obstacle detection with the help of CMOS digital image sensor for short term memory before finalizing decision to avoid the obstacle, with its simulation tabulated as variation of coordinates keeping internal parameters as predefined by manufacturer.
Two-dimensional lens modeling [2, 3] Information from the lens is the input to all vision-based systems, hence, it is obligatory to be faithful and accurate. Therefore, modeling of lenses needs to be handle more carefully. Figure 1 shows 2-D lens modeling perspective projection. The image formation i
Data Loading...
