Image analysis can be defined as the complex of operations of image recording, image processing, and identifying, measuring and classifying objects and shapes in an image. An image can be a photograph, drawing, or video. Furthermore, an image can be a single greylevel image (e.g. X-ray or thermal infrared), a colour image, or a spectral image (many different bands). Objects can be seen at different scales, using the range from microscopes to satellites.  

Biometris focusses on digital still images for agricultural purposes, mostly regarding plant material. Topics on image analysis include:

• Measurement of size and shape of plant parts
• FLORES image database: identifying varieties of roses by comparing images of flowers
• Monitoring stress of leaves of plants in time, using the MIPS-facility
• Measuring compounds in fruits and seeds using hyperspectral imaging
• Measuring grass land using hyperspectral imaging and remote sensing

For vision and imaging projects we are part of GreenVision, the expertise centre for image processing in the agro- and food industry.
Biometris also participates in the Wageningen UR fieldrobot team 'Sietse'.

• GreenVision website
• Fieldrobot 'Sietse'
  
  

Measurement of size and shape of plant parts

Over the years, we have developed a range of routines to measure the size (length, width, area, volume) of a plant or plant part. Also the shape can be measured. Simple shape features like length/width ratio often work well, but more complex Fourier descriptors are also informative. A major research theme was focused on shape similarity of objects. 

        
An example of average shape comparison of carrots. The average shape of 10 different varieties in 28 plots are shown. Clearly the difference between varieties can be observed. With image analysis, this difference can be quantified and used as an objective, repeatable measure. 

Simple size and shape measurements on flax seeds.