| This histogram can display the area per color class or group as well as the object frequency distribution in function of a morphological measurement (Ex: number of objects of a certain area or length range). |
|
|
|
|
 |
| The original image to analyse is a 24 bits color image which can contain millions of colors (it is not limited to 256 as in some competing programs). Images in such format are easily obtained with current digital cameras and scanners. This image contains all the scene fine color variations that the human eye is accustomed to. Note: The image can also be a 256 grey levels image. |
Before analysing a series of images, you must teach WinCAM which colors are part of the objects to analyse (the grass, fruits, disease or healthy regions) and which ones are not (the background such as the soil) simply by clicking the image and giving names to these clusters that we call color classes. In the first phase of the analysis, WinCAM will classify the image pixels into one of these classes that you have defined (special processing is provided for those who do no match any criteria). The result of this classification can be displayed (see above and below images) and saved to tiff files. |
Color classes can be grouped, the group can be given a name and area measurement data can be obtained per color class or group. In the above example, the Grass group is made of four green colors and the soil two brown colors.
The classification into groups can also be displayed (see above) and saved to tiff files. There can be up to three groups and each one can have up to twelve color classes. |
Example of color classes to analyse the soil cover image to the left.
|
|
|
|
|
Example of color classes to analyse the image to the left.
|
| Pixels are classified as a function of their color (Hue, Saturation and Intensity which is equivalent to Hue, Value and Chroma used in the Munsell color classification system). A tolerance factor for each of these component allow to include more or less colors in each class. |
| Pixels classification into Hue, Saturation and Intensity can also be displayed on screen and exported to tiff files. |
Color image
|
|
| One of these channels, as well as the Red, Green or Blue channel can also be displayed and analysed instead of the original color image. |
Red channel
|
Green channel
|
Blue channel
|
Hue
|
Saturation
|
Intensity
|
|
|
Measurements
|
|
In the first phase of the analysis, colors or grey levels are used to differentiate objects from their surrounding background. Debris can be filtered out by different methods (see table below). In the second analysis phase, measurements are made on the detected objects and the whole image. These fall into four categories.
- 1) Global measurements related to the analysed region such as total and percentage area occupied by objects and their average morphological measurements (area, length, width...).
- 2) Individual morphological measurements such as area, length, width, perimeter and form ceofficient.
- 3) Area in function of color. For each object and the analysed region, you get the area occupied by each color of interest and groups of them. This also includes information about the dominant color.
- 4) Interactive measurements. To manually or semi-automatically measure objects dimensions by clicking them in the image.
|
Note:
You can get interactive information about analysed objects by clicking them in the image or look at the data files for more details. Measurements data are saved in text files that can be read by most data analysis or spreadsheet style programs (like Excel).
WinCAM can acquire images directly from most optical scanners or cameras that have a TWAIN driver and can also analyse images from almost any digital camera stored in tiff or jpeg files.
|
|
Color calibration
|
| When you classify soils or plants by comparing their color to a printed set of reference colors such as the Munsell Soil Color Charts or Munsell Plant Tissue Color Chart, you are doing what we call an absolute classification. When you classify plant colors without comparison to an external reference, simply by comparing them (such as when determining which ones are darker green and which one have a yellowish tint), you are doing what we call a relative color analysis. WinCAM works in the latter way, that is it classifies colors on a relative basis of what is contained in an image. It is possible in WinCAM to emulate an absolute color analysis by including in each image some reference colors (such as a subset of the Munsell samples). In that case, colors of the objects to analyse are compared to these reference targets. This will work for as long as the number of reference classes do not exceed WinCAM's limit of 36. Also, in order to minise measurement differences among images which do not have color references, it is possible to calibrate the color balance against a white object (such as the targets which come our Field Image Acquisition Kit illustrated on back cover). |
|
WinCAM differs from the Regular Version by its ability to analyse individual objects
|
The Pro version is required to analyse individual rice kernels morphology or to classify them in function of color (colors distributuion per kernel).
The analysis of soil covered by plants as illustrated above can be done with the Regular or Pro versions. There are no objects to analyse in such images (only the classification of pixels into two groups, each having one or more colors). With both versions you can analyse the whole image or simply outline the region you wish to be analysed |
 |
|
WinCAM is available in two versions
(Table below lists their respective features and measurements)
|
|
Features/Measurements
|
Reg
|
Pro
|
Description
|
| Analyse whole image or sub-region of it |
Y
|
Y
|
Measure areas in function of color for whole image or a sub-region of it of any shape. Rectangular and circular are pre-defined shapes. Create any irregular shape regions with the lasso tool (outline their boundary by drawing in the image). You can move and resize regions which have been analysed. |
| Individual objects morphology and color analysis |
N
|
Y
|
Analyse non-touching objects morphology and color on an indivitual basis. Measure the objects length by 4 methods, width, perimeter, form coefficient, length/width ratio, object area, area per color and information about dominant color. Objects presence and how measurements have been done on them can be indicated by different symbols over the image. |
| Debris filtering based on morphology |
N
|
Y
|
Remove debris based on anatomical properties like area, length, width, length/width ratio and form coefficient. |
| Debris filtering based on color |
Y
|
Y
|
When debiris have a different color than the objects to analye, this function can be used to exclude them from the analysis |
| Analyse grey levels or color images |
Y
|
Y
|
Analyse the colors of a color image or the luminosity in grey levels image. Color or grey level information can be used to separate pixels belonging to objects and background. |
| View/Analyse/Export RGB channels |
Y
|
Y
|
From a clor image, choose its Red, Green or Blue component to be displayed on screen, analysed and saved to a standard tiff file. |
| View/Analyse/Export HSI channels |
|
|
From a color image, choose its Hue, Saturation or Intensity color information channel to be displayed on screen, analysed and saved to a stardard tiff file. The HSI components are equivalent to Munsell's Hue, Value and Chroma components. |
| Exclusion regions |
Y
|
Y
|
Exclude from the analysis any sub-region of any shape of the image. Can be used to remove debris or to avoid areas which must not be analysed. |
| Batch analysis |
Y
|
Y
|
Analyse images without operator supervision. Note: Images which require operator intervention for calibration canno be analysed in batch. |
| Interactive pixels classification adjustment |
Y
|
Y
|
Allows to make modifications to the automatic pixels calssification (into object and background) based on color or grey levels in specific regions or the image (any shape) or the whole image. |
| Analysis can be stopped |
Y
|
Y
|
Simply press a keyboard key while the analysis progress bar is displayed. |
| Keep or reject incomplete objects |
Y
|
Y
|
Prevent objects truncated by image boundaries from being analysed and affecting average measurements. |
| , |
| Pixel size calibration |
Y
|
Y
|
For images from camera, pixels size must be determined by calibration to produce measurements in units of your choice (mm, cm, inches...). With Regent's targets (see Field Image Acquisition Kit below), all is needed is a mouse click on the target in the image. |
| White balance calibration |
Y
|
Y
|
To remove color variations due to changes in lighting or camera settings between images. With Regent's targets (see Field Image Acquisition Kit below), all is needed is a mouse click on the target in the image. |
| Remove background light variations |
Y
|
Y
|
For images acquired with a camera when background light has variations which can affects the analysis. Note: It is best to invest time in having a uniform lighting. The analysis will be faster and easier. |
| , |
| Image edition |
Y
|
Y
|
To remove debris or artefacts. Can also be used to make corrections to pixels classification (into object and background), to trace over objects to analyse (when this is not possible automatically) or to separate touching objects. You can edit with predefined colors or pick those already present in the image. |
| Interactive measurements/Number of methods |
Y/1
|
Y/2
|
Interactive measurement is a way of measuring objects manually (both versions) or semi-automatically (Pro version) when the automatic analysis cannot do exactly what you want to measure. In semi-automatic mode, simply draw a line over an object at the place you want it to be measured and WinCAM will automatically find the object's boundary beginning and ending over that line and measure the length in between (it will also measure the travelled distance). |
| Histogram of frequency distribution |
N
|
Y
|
Display the number of objects that fall into classes of morphological features (Ex: number of objects with an area between 1 and 2 square mm). |
| Histogram of color distribution |
Y
|
Y
|
Display the area covered by each color galss or group in the analysed region or image. |
| WinCAM comes with an instruction manual with plenty of color illustrations, examples of applications and images for training. It also comes with prompt technical support (by email exclusively) by Regent Instruments personnel (the answering time is typically one hour when messages are received during North America eastern time working hours). |
|
Field Image Acquisition Kit
|
 |
The field image acquisition kit is for in-field images acquisition with digital cameras for their analysis with Regent's programs. One of its component is a white plastic backgroung (available in 4 different sizes) with a translucent anti-reflective fold-up cover that presses thin objects against it so that they lay flat. The kit comes with stick-on waterproof removable calibration targets which Regent's programs uses to balance the colors and measure objects dimensions. You enter their dimensions once and calibrating an image acquired with them involves a single mouse click on the target. An optional item is the tripod extender which allows to position the camera over the centre of the white background (away from the tripod legs). |
