Development of a tool to analyze back-focus characteristics of fix-focal cameras

Abstract

Fix-focal cameras require careful alignment of the sensor relative to the lens during assembly in order to produce a sharp image. Axis developed wide- and fish-eye network cameras are, for a number of reasons, focused at infinity when aligned.

This thesis develops the software for a new inhouse tool to be used at Axis to analyze the back-focus effect on the camera’s focus characteristics. The tool is a machine called Modus, which uses a variable focus collimator to measure the sharpness of objects at distances not limited to infinity.

The main purpose of this new tool is to answer the following questions:

• How can knowledge of the back-focus effects for a given camera be harnessed to decide on back-focus offsets to achieve desired focus characteristics during alignment?

• What are the sensor placement tolerances to achieve a certain image sharpness? Can this information be used to decide between active or passive alignment?

A fish-eye camera, Koi, is analyzed with the new software written for Modus. The
results show that the focus moves from infinity to 0.3 m within a back-focus window of 5 µm, with focus changes of ± 0-5%. The depth of focus for a number of image sharpness levels is also measured.

The limitation of the applicability of the results for the Koi camera lie in the sample size. In this thesis only one unit of the camera is analyzed. In order to draw conclusions about appropriate back-focus offsets a larger sample size in needed.

Abstract (Swedish)

Vid sammansättningen av kameror ställs höga krav på finjustering av sensorn och
objektivets relativa position för att den resulterande bilden ska bli skarp. Axisutvecklade nätverkskameror fokuseras, av olika anledningar, i oändligheten.

Examensarbetet utvecklar mjukvara för ett nytt verktyg, Modus, som ska användas
in-house för att undersöka effekten av kamerans backfokus på bildens skärpa.
Verktyget är en maskin som är utrustad med en justerbar kollimator som gör det
möjligt att mäta skärpan av objekt som även är närmare än oändligheten.

Huvudsyftet med det nya verktyget är att kunna samla in data som ska besvara de
följande frågorna:

• Hur kan information om effekterna av backfokus användas för att bestämma lämpliga backfokus-offsets ute i produktion för att på så vis uppnå önskvärd bildskärpa för en kamera vid olika objektavstånd?

• Vad är toleransen för placering av sensorn för att uppnå en specifik bildskärpa? Kan denna information användas för att bestämma om en produkt ska sammanfogas med aktiv eller passiv alignment?

En fish-eye kamera, Koi, analyseras med det nyutvecklade Modusverktyget.
Resultaten visar att fokus skiftar från oändligheten till 0.3 m då backfokus ändras med 5 µm. Skärpan förändras ± 0-5 % för alla objektavstånd. Toleransen mäts också upp vid ett antal olika nivåer av bildskärpa.

Begränsningen av resultatets användning ligger främst i provstorleken. Endast en
enhet av Koi-kameran har analyserats. För att kunna uttala sig om lämpliga
backfokus-offsets behövs analyser utav fler enheter.

Popular Abstract

Enabling analysis of Axis cameras failing optical tests in production

A hundred percent yield in the production of any product is a desirable but often unrealistic goal. Imperfect assembly lines, operator errors or faulty components are a reality, and so also in the production of Axis network cameras. The important thing is to be able to pinpoint what went wrong to attempt to rectify or at the very least monitor it. This thesis has developed an inhouse tool for Axis to analyze how the focus has shifted in cameras that failed optical tests in production – as well as analyze the cameras at an earlier stage, during development, to find out more about their optical characteristics.

The covid-19 pandemic has made component shortage a phrase heard across news channels and a reality for many companies. Axis Communications produces network cameras mainly used for surveillance and has not been spared. The difficulty in keeping a large stock of spare parts has made it more important to not waste the parts that are available. However, production invariably has some drop-out with yields rarely at one hundred percent. The failed units can be caused by many different things such as errors in the handling of components, faulty components, inherent systematic errors in the production line et cetera. Analyzing why units fail in production is key to continuous improvement of production lines and machines.

One of the first steps in the production of Axis network cameras is the alignment of the image sensor and lens. The goal is to position these relative to each other such that a sharp image is achieved. This might sound trivial because many of the cameras have autofocus, but the orientation of the sensor relative the lens must be considered. A misalignment of as little as 0.02° can be enough to cause different parts of the image to be in focus at very different distances. Additionally, shifting focus is undesirable in the application of some Axis cameras. As such the alignment step becomes the only time the focus is set and thus very crucial.

This thesis has developed an inhouse measurement system for Axis network cameras that can assist in understanding why cameras fail optical focus tests in production. It enables analysis of the focus of partially assembled cameras as well as more comprehensive mapping of focus characteristics of units during development of a camera. The focus is quantified and can be measured at objects anywhere from a few decimeters away all the way to infinity (stars in the night sky) using an optical instrument called a variable focus collimator.

The measurement system can also analyze the focus characteristics of the camera prior to alignment when the image sensor and lens can move freely relative to each other. This can be used during the development of new cameras to analyze how much a small translation of the sensor relative to the lens shifts the focus. Measurements performed on a fish-eye camera showed that translating the sensor 5 micrometers away from the lens results in the focus shifting from infinity to 300 mm. This
information can be used to determine which offset should be aimed for in the alignment process to set the focus at a certain distance.

The work focused mainly on developing the software for the machine. Programs were written that autonomously scans through several sensor positions, object distances, incidence angles and measures the focus at each position. The result is presented both graphically as well as in raw data files that can be used for further analysis. The system is used regularly inhouse by engineers and with further development has the potential of being used at factories during production.

• author: Lind, Björn
• supervisor: Per-Erik Andersson
• organization: Innovation
• alternative title: Utvecklingen av ett verktyg för att analysera backfokuskarakteristik för fixfokal-kameror
• course: MMKM05 20192
• year: 2022
• type: H2 - Master's Degree (Two Years)
• subject: Technology and Engineering
• language: English
• id: 9075412
• date added to LUP: 2022-02-16 07:37:05
• date last changed: 2022-02-16 07:37:05