IMAGE metrics

Image Metrics is an image analysis application. It performs analysis of digital images produced from Digital Camera Modules/Image Sensors. Image Metrics contains a suite of applications to support image analysis. "Image Metrics" outputs analysis reports and auxiliary measurement data that is usable by image scientists and test engineers for improving image processing pipelines and manufacturing processes. The Image Metrics user interface is intuitive and easy to use, with logical navigation and many features and tools to aid in the investigation of image quality issues. Image Metrics contains many of its own analysis algorithms but also contains SMIA and CPIQ algorithms. MIP Systems can develop customer specific analysis algorithms and be added to Image Metrics and the Software Development Kit.

main functions

Analysis List

All of the analysis algorithms are listed and selectable from a single list for execution with parameters for each to modify to your specifications. Image Metrics covers a broad range of algorithms from noise to color, sharpness and contamination. Each analysis algorithm has default parameters and a short description of the algorithm.

Color Conversion

Image Metrics can perform image color conversions for RAW images, and standard RGB images to other image planes before executing an analysis algorithm. Conversions for RAW images are "Extract Bayer" planes, Auto White Balance (merges Bayer after AWB), "Decode Bayer", "Merge RGB", "Extract YUV", "Extract HSL", and for standard RGB images, "Extract RGB" planes, and "Extract YUV". Image Metrics also supports image averaging by opening a directory of images and selecting "Average Images" from file menu to generate an averaged image.

Measurement Palettes

The Measurement palettes are a powerful toolset to display image parameteric values using the whole image or a drawn Region Of Interest selection. Image Metrics contains serveral Measurement palettes, including a Histgram palette, Intensiy Map, 3D Map, Sharpness palette, and a Matrix palette. A Sharpness palette shows the Sharpness score from a ROI drawn over an objects edge. A Matrix palette shows a set of ROI's values using the Matrix tool, and can be viewd in the Intensity Map and 3D Map palettes. All palettes are updated when you change the color conversion settings or open or select another image file. All values shown on palettes (except 3D Map) can be exported into a text file for further study.

Settings Palettes

Image Metrics contains serveral Settings palettes. The RGB Settings panel adjust the RGB image values by balance percentages, and the White Balance Settings panel adjust the Bayer planes of a RAW image. The Lab Settings panel contains the Lab conversion parameters. The Math Settings panel is a math step sequence editor to perform addtional computations on image data before running any analysis functions. The Matrix ROI settings panel contains the parameters for adjusting the Matrix of ROI's size and spacing. The RAW File Settings panel is an editor for RAW image files parameters, you can edit and create new parameters; which can be exported and imported to share with others. If your RAW image does not look correct you can edit or select another parameter set and the image file will automatically be reloaded using new parameters.

Analysis results

Analysis Reports

All Analysis algorithms output an Analysis Report with one or more data sheets. These data sheets contain overall test results and auxillary data derived from the alorithms internal steps. These data sheets can be individually exported to a text file or all data sheets to a single text file. The Analysis Report sheets can be very useful for investigating image quality issues, and correlating with customer designed algorithms.

Analysis Graphs

Some of the Analysis Algorithms may output Analysis Graphs. These graphs represent data processed from steps within the algorithm. These graphs can be exported to a CSV file for further study in your favorite spreadsheet application. These Analysis Graphs can be very useful for investigating image quality issues, and correlating with customer designed algorithms.

Processed Images

Many of the Analysis Algorithms output Processed images. These images may contain overlays indicating where measurements where made and their values. Some of these images may be processed or filtered images from steps within the Analysis algorithm. These imagess can be exported to an image file. These processed images can be very useful for investigating image quality issues, and correlating with customer designed algorithms.

histogram The Histogram palette quantifies the entire image or quantifies from a Region of Interest tool. This is very useful function as it gives you a snap shot look at the grouping or binning of pixel values. You can change the bin count and the minimum and maximum values to isolate a particular group of pixel values. This function allows exporting of the histogram data to MS-Excel or a simple tab-delimited text file.

intensity map The 2D Intensity Map palette displays pixel data from a Region of Interest tool. This is very useful function as it gives you a subset view of image data or a regional matrix view of data from intensity maps with the option to select color palettes. This data can be exported as a CSV data file.

3D Map The 3D Map Palette displays pixel data from a Region of Interest tool. This is very useful function as it gives you a 3-dimensional view of image data to understand the contours of the image. You can "Render Window” the 3D Map into its own window so you can manipulate the rendered window for optimal viewing on another monitor as you select other data maps or images.

sharpness The Sharpness Palette displays the MTF curve from a Region of Interest tool. This is very useful function for a quick check of slanted edges in your image. You can run the Sharpness algorithm using a template file for a more inclusive look at more slanted edges in the image. This data can be exported to a file.

matrix This Matrix Palette shows the a set of predefined measuresments for the Matrix ROI tool. You can select from five measurement types, Mean, Maximum, Minimum, STDDEV, and SNR. The values are shown in a spreadsheet indicator and can be exported to a text file for further study.

RGB Color Balance The RGB Balance Settings let you adjust the Red, Green, or Blue values by percentages for a loaded RGB image. This can be useful to measure the amount of color offset in one of the channels.

bayer white balance The White Balance Settings allows to set the ROI size in the center of the image to perform the White Balance on RAW Bayer images. This levels all 4 Bayer planes to the same means for a White Balanced image.

LENS SHADING CORRECTION Corrects the lens shading effect in an image by applying correction factors using Relative Illumination percentage. You can input a RI value or enable the Relative Illumination function to measure the RI value and apply it to the LSC algorithm function to correct for the lens shading effect.

lab settings The Lab Settings allows you to modify the RGB to Lab conversion parameters. This is performed when L*, a*, or b* color plane has been selected from the "Color Conversion Toolbar".

math editor The Math Editor performs arithmatic on an image before executing analysis algorithms. This is done by creating a Math Step file and running the file on the selected image. This function can be very powerful using four functions, plus, minus, multiply, divide and using a constant, Image Minimum, Maximum, Mean, Previous Step Value, as the Numerator or Denominator. You can create many math scripts and are saved locally to the Image Metrics database.

matrix settings The Matrix Settings allows yout to change the number, size, and spacing of the ROI's conatianed within the drawn ROI. You can apply your changes and all measurements are updated automatically.

raw file settings The RAW File settings are used decode the RAW file for proper display and use. Image Metrics allows you set RAW File settings and save it to Image Metrics local database, and you can also export the settings file to an INI file for use by the Read Image File function in the SDK.

Alignment Measures XY Tilt, Rotation and other alignment measurements.

Color Chart Determines whether the Color from the 24 color patches are within an acceptable range, by comparing their L*a*b* values to a reference file.

Color Ratio Determine whether certain Color Ratios fall outside acceptable limits at predefined locations.

Color Uniformity Radial Determines color shading irregularities across an image using ROIs in a ring orientation (multiple rings are used across image).

CPIQ-Color Shading Determines shading irregularities in intensity and color across an image using a grid of ROIs.

Dark Noise Determine whether the image sensors temporal noise in a Dark Field is within acceptable range. Dark Noise is calculated by using 2 RAW images that were consecutively captured from the image sensor.

Data Line Integrity Determine whether the image sensor has data lines that are either stuck low, high, or stuck to adjacent data lines. This is evaluated by subtracting the self test pattern image from a Reference image with the same test pattern and the percentage of bad pixels is calculated.

Defective Pixels Scans image for contamination particles. A particle is characterized as a pixel stuck low, a pixel stuck high, or a pixel covered by staining or foreign material at any level of the module. This algorithm detects defects or contamination, and uses an inner and border region mask to bin defects. Inner region defects can have more strict pass fail criteria and a more lenient criteria for the border region. This algorithm reports three types of particle sizes; Single pixel, Couplets (2 adjacent pixels), and Clusters (3 or more adjacent pixels).

Defective Pixel Pairs Scans image for defective pixels. A defective pixel is characterized as a pixel stuck low, a pixel stuck high, or a pixel covered by staining or foreign material at any level of the module. Single Pixel (SP) defects are detected and four Defect Pair types are determined within a 5x5 pixel grid of the Single Pixel defect, ARPD, SPD, MPD, LPD.

Geometric Distortion Determines whether the imaging optics has excessive barrel or pincushion distortion. A target chart with a matrix of dots is required for this algorithm.

Line Noise Determine whether the image sensor is creating temporal noise or fixed pattern noise. Line Noise is calculated by using 2 RAW images that were consecutively captured from the image sensor.

Low Contrast Defect Scans image for low contrast defects. A low contrast defect is characterized as a colored or gray area in the image, which may represent a water spot, residue on sensor surface, or particles on the IRCF, Lens or CFA abnormalities.

Optical Center Determines whether the imaging optics is centered with respect to the image sensor. The measured results are relative pixels values from image center. An uniform light field image should be used.

Relative Illumination Determines whether the imaging optics has excessive lens shading roll off. Relative illumination is calculated by dividing the average of each ROI in each corner of the image (upper left UL, upper right UR, lower left LL, lower right LR) by the average value of the ROI at the image center. Relative Illumination delta is the difference between the brightest and the dimmest corner.

Relative Uniformity Determines determines whether the imaging optics has excessive lens shading roll off, contamination, or irregular luminance values across the image. Relative Uniformity is calculated by dividing the image into a matrix of ROIs and calculating their means and then the maximum deviation from its neighbors and reporting the maximum deviation for the corner region set by “Corner Blocks” and the maximum for edges, and center regions.

Sharpness Determine whether the slanted edges specified by the template file are within acceptable range. The focus scores can be specified as SFR, MTF, CTF, or CTF with W/B reference patches. Delta values can be specified in any measurement group, by using the Template Designer.

SMIA-Blemish Scans image for blemishes. A blemish is characterized as a colored or gray area in the image, which may represent a water spot, residue on sensor surface, or particles on the IRCF, Lens or CFA abnormalities. Blemishes are defined into two classifications, Minor and Major. A Minor defect is defined as a single blemish defect with no adjacent defects, and a Major defect is defined as more than one adjacent blemish defect.

SMIA-Defective Pixels Scans image for defective pixels. A defective pixel is characterized as a pixel stuck low, a pixel stuck high, or a pixel covered by staining or foreign material at any level of the module. Multiple thresholds are used to determine weak or dead pixels and further classify them into minor and major couplets in a inner and border region. Clusters are classified as 3 or more defect pixels of any kind.

SMIA-Dynamic Range The dynamic range of a camera module is a measure of the range of light levels that may be present within one scene and reproduced faithfully. The upper useable limit of the light response of the camera is termed the full-scale deflection (FSD) of the camera. The minimum discernable response is taken to be at one standard deviation of the noise, including dark noise, above the noise floor.

SMIA-Fixed pattern Noise Fixed Pattern Noise is calculated by finding the row and column averages and calculating the variation between column averages (VFPN) and row averages (HFPN). If Multiple Images are selected then images are averaged before analysis.

SMIA-Row Column Noise The Row and Column Noise for an image camera is a measure of the temporal noise present in row averages and column averages, which manifests itself as flickering rows or columns when imaging in low light conditions.

SMIA-Signal Noise Ratio The signal to noise ratio (SNR light) for a camera module is a measure of the amount of speckle in an image of a lit scene. The SNR light can be defined as a noise power level for a standard uniform illumination, which, along with an exposure, results in an average output of 50 ± 5% of the FSD. As with the dark temporal noise, the SNR light is measured by taking a number of frames and finding the mean and standard deviation of the pixel level over these frames for each pixel. The pixel standard deviations represents the noise in an individual pixel, which are then root mean squared and divided into the means to give the SNR value for the camera.

SMIA-Temporal Noise The temporal noise is a measure of the “speckle” component of an image. The temporal noise is seen as a pixel level that varies randomly from frame to frame. The temporal noise is measured by taking a number of frames and finding the standard deviation of the pixel level over these frames for each pixel. The pixel standard deviations represents the noise in an individual pixel, which are then root mean squared to give a temporal noise value for the camera.

Step Chart Determines whether the Color from the step chart patches are within an acceptable range, by comparing their L*a*b* values to a reference file. The color values are measured by calculating the average pixel value of R, G, and B within each color patch, and then the L*a*b* values are calculated. The delta between the measured L*a*b* values and the Reference File L*a*b* values are calculated and then reported.

Veiling Glare Determines whether the Veiling Glare Index is within acceptable range. Veiling glare is the reduction in contrast, or misting, in an optical system due to random scattering of light onto the image plane. The veiling glare index is defined as the ratio of the irradiance at the center of an image of a small perfectly black area superimposed on an extended field of uniform radiance, to the irradiance at the same point of the image plane when the black area is removed.

file and data types

File Types: Image Metrics supports the following image file types; BIN, BMP, IMS, IMT, JPG, PNG, RAW. The IMS and IMT file types are specific to the Image Metrics application. BIN and RAW file types are considered a RAW image format. RAW image files require a "RAW File Settings" to open.

data Types: Image Metrics supports image data with bit depths from 8-bit to 64-bit, BAYER interpolation to RGB format from 8-bit (U32), floating point image data via CSV files, and all image data calculations are performed in double precision representation.

Computer requirements

Item Minimum Recommended

Processor Duo Core or equivalent i7 Quad Core or greater

RAM 8GB 16 GB or greater

Screen Resolution 1440 x 900 pixels 1920 x 1080 pixels or greater

Operating System Windows 8/7 Pro
(64-bit)
Windows 10 Pro (64-bit)

Disk Space 100 MB 100 MB

USB License Key

NONE Required

Specifications