Join Imatest at the Electronic Imaging Symposium from Jan. 13 – 17 in Burlingame, California USA. Explore the entire imaging science ecosystem, from capture through image processing to how we and our surrogate machines see and interpret images. Henry Koren and Norman Koren will be presenting on the following.
Reducing the cross-lab variation of image quality metrics
Henry Koren
As imaging test labs seek to obtain objective performance scores of camera systems, many factors can skew the results. IEEE Camera Phone Image Quality (CPIQ) Conformity Assessment Steering Committee (CASC) working group members performed round-robin studies where an assortment of mobile devices was tested within heterogeneous imaging labs. This paper investigates how the existence of near-infrared energy in light sources that attempt to simulate CIE illuminants can influence test results. Numerous other impacts, including the influence of opal diffusers used for uniformity testing, how test scene framing can alter white balance and exposure, and how chart quality and texture frequency distribution can skew results. We introduce a test procedure which is intended to reduce intra-lab variability and a method for assessing an independent lab’s competence in conforming with the IEEE testing standards.
Date/Time: Tuesday January 15, 2019, 3:50 – 4:10
Location: Grand Peninsula Ballroom E
Compensating MTF measurements for Chart Quality limitations
Norman Koren
Objective measurements of imaging system sharpness are typically derived from test chart images. It is generally assumed that if testing instructions are followed—if the chart print quality is fine enough and the chart magnification (on the sensor) is low enough— test chart quality will have little impact on the overall measurement.
This assumption may not be valid when extremely high-resolution cameras are tested with standard charts or when smaller than optimum test charts are used because of laboratory or production line space limitations.
MTF compensation is applied by dividing the measured system MTF by chart-projected MTF as a function of the sensor.
- Measurements made under different conditions— with different test charts (transmissive as well as reflective) by different people in different labs— are more consistent.
- The megapixel suitability of test charts is increased by a factor of approximately 2 (1.4x linearly).
- The key advantage of MTF compensation is that measurements are more accurate (in an absolute sense).
Date/Time: Tuesday January 15, 2019, 9:10am
Location: Grand Peninsula Ballroom E