[PDF] FPGA ARCHITECTURE FOR REAL-TIME BARREL DISTORTION PDF ICME2011.pdf

Index Terms— real-time barrel distortion correction, frame buffer, Field Programmable Gate Array (FPGA), YUV 4:2:2 1 INTRODUCTION In recent years digital

ABSTRACT Images acquired by a fisheye lens camera contain geometric distortion that results in defor- mation of the object's shape To correct the lens distortion,

An Efficient Barrel Distortion Correction Processor for - IEEE Xplore

19 jui 2018 · can perform the barrel distortion correction (BDC) jointly with color demosaicking, so as to produce the barrel-distortion-corrected color image

[PDF] FPGA ARCHITECTURE FOR REAL-TIME BARREL DISTORTION

Index Terms— real-time barrel distortion correction, frame buffer, Field Programmable Gate Array (FPGA), YUV 4:2:2 1 INTRODUCTION In recent years digital

[PDF] A Real-time FPGA Implementation of a Barrel Distortion Correction

A higher quality lens can be used to correct for this but this comes at additional cost to the image capture system Barrel distortion is primarily radial in nature, with

[PDF] A New Approach to Lens Distortion Correction - Massey University

When using an imaging system to make measurements, it is important to correct for lens distortion An image of a grid is captured, and parabolas are fitted to

[PDF] Automatic Radial Distortion Estimation from a Single Image

tion · Radial distortion · Barrel distortion · Pincushion distortion · Plumb-line sumption is invalid, making distortion correction a must Radial distortion is the

COLOUR IMAGES

Henryk Blasinski

ECE Deptartment

University of Rochester, NY

henryk.blasinski@rochester.eduWei Hai , Frantz Lohier

Logitech Inc.

6505 Kaiser Drive

Fremont, CA

ABSTRACT

This paper presents a hardware architecture for real time bar- rel distortion correction of YUV 4:2:2 encoded color images. In our solution we are presenting an alternative implementa- tion of the correction engine which is based on multiplica- tionsrather than trigonometrictransforms. The system makes minimaluse of resourcesthus being a good candidate for em- bedding into a single chip. Proposed solution is implemented in a cost-effective Spartan 3 Field Programmable Gate Array (FPGA). Our architecture iscapable of processing QQVGA images at the rate of 30 frames per second (fps). Qualitative and quantitative examination confirm correct preservation of color information in processed images. Index Termsreal-time barrel distortion correction, frame buffer, Field Programmable Gate Array (FPGA), YUV 4:2:2

1. INTRODUCTION

In recent years digital camerashave become omnipresent in our daily lives either as stand alone devices or embedded into mobile phones, cars (back-upcameras) or surgical equipment (laparoscopes). Typically such cameras have a field of view (FOV) of about60◦ , which is only a third of the FOV of the human visual system. On the other hand there exist lenses coveringmuchlargerFOVs, suchaswide-angle(120 -130 or fish-eye (≂180 ). They exhibit, however, a significant amount of barrel distortion, making them less appropriate for high volume, consumer applications. Large FOV optics is present in certain niche markets including: automotive, med- ical and surveillance industries [1]. eral correction algorithms have been proposed [1-3]. Also in recent years some work has been done on hardware imple- mentations of those algorithms, however, authors focused on functionalityratherthanresourceconstraints[4-8]. Moreover only grayscale images have been taken under consideration, significantly reducing their impact on consumer applications.? Wei Hai is currently with Contour Inc.; wei@contour.com Frantz Lohier is currently with Kudelski Group; frantz@lohier.com In this paper we are presenting an FPGA architecture for real-time barrel distortion correction of colour images en- coded in the YUV 4:2:2 format which is frequently used in the industry [9]. In addition to this we are proposing an al- ternative implementation of the correction engine, allowing for more efficient utilisation of resources available on modern

FPGAs.

2. BARREL DISTORTION CORRECTION

Barrel distortion is one of the most common imperfection of lenses, it causes straight lines from the real world to be repre- sented as curvesin the image. It can be regardedas a geomet- ric image transformation. It converts an undistorted imageI with pixel coordinates(x,y)into a distorted imageI dwith coordinates(x d ,y d ). The goal of the distortion correction is to perform an inverse operation that is to find a corrected im- ageI c with coordinates(x c ,y c )such thatI c ≈I.

2.1. The polynominal method

According to [10] the barrel distortion can be decomposed into two primary components, tangential and radial. The ra- dial component is affecting the distanceRbetween any point in the image and the optical centre. The tangential compo- nent is affecting the angleθbetween the line joining given pointand the projection centre. The relationship between dis- torted valuesRd andθ d and corrected onesR c andθ c is best described by a polynominal of then th order: c n k=0 a k kd =p(θ d )(1) R c n? k=0 b k ·R kd =q(R d )(2) Lenses can be considered symmetric with respect to their op- tical centre. This means thata 0 =b 0 =0and the tangential distortion is negligible:θ c d , making the radial distor- tion the only component needed to be accounted for. Such assumptions have been made in many earlier works [3-8] Hardware implementation of the barrel distortion correc- tion algorithm can be done in several ways. The simplest approach is to explicitly encode the relationship between all distorted and corrected image coordinatesin a form of a large look-up table [5]. This system may be very fast, however forlarger image sizes it has significantmemoryrequirements.

Thesecondapproachusesa CORDICenginetoperformpolar

to rectangular conversion and implements polynominal map- ping between distorted and corrected radii, as described by (2) [7,8]. In this case, the main constraint is the FPGA chip size, since CORDIC engine requires significant amount of logic.

2.2. The coefficient method

The major disadvantage of the polynominalmethod is the ne- cessity of performingrectangular to polar and polar to rectan- gulartransforms[8]. We wouldlike toproposea differentfor- mulation of the polynominal method allowing to avoid those transitions. It is based on the assumption that the tangential distortion is negligible (θ c d The polynominal method describes the relationship be- tween a point in the disorted domain and its equivalent in the corrected domain:(x d ,y d )→(x c ,y c ). For hardware ap- plications it is much more convenient to perform an inverse operation:(x c ,y c )→(x d ,y d )that is for a particular point in the corrected image,calculate its coordinates in the distorted image space. This approach,called the back mapping [7], relates radii as:R d =m(R c ). In many applications the func- tionm(R c )is also approximatedbypolynominals. Addition- ally the back mapping is more convenient for implementing image interpolation techniques [7]. From the congruence of coordinate triangles (Fig. 1) in both domains we may write: R d x d =R c x c (3) R d y d =R c y c (4) after some rearrangements one obtains: x d =x c ·R d R c =x c ·C f (Rc,R d (5) y d =y c ·R d R c =y c ·C f (Rc,R d (6)

Aboveequationsrevealthat in orderto obtain(x

d ,y d )co- ordinatesin the distorted image space it is enough to multiply both undistorted image coordinates(x c ,y c )by the same co- efficientC fquotesdbs_dbs17.pdfusesText_23

[PDF] [PDF] FPGA ARCHITECTURE FOR REAL-TIME BARREL DISTORTION

[PDF] Correction of Barrel Distortion in Fisheye Lens Images Using Image