Best practices for completing a successful scan

When you’re just getting familiarized with your new Harvest Quality Vision (HQV) scanning system, it’s normal to need to practice a little in order to take an effective scan sample. The following includes key practices to follow for a high quality scan, and the solutions to common problems that may come up.

• Calibrate your structure sensor camera for HQV scanning. Camera calibration will only need to be completed upon initial setup, or if the camera has physically shifted over time, or has been moved. Calibration requires good lighting – natural light is ideal. A full video demonstration of how to get started on calibrating the sensor camera can be found here.
  
  If the colour of the side of the bin, table or floor appears to bleed onto the apples in a scan image, this means the colour capture is not aligned with the 3D sensor. This requires re-calibration.
  
  
  
  
  
  

• Allow iPad time to adjust for white balance. In some instances, the color accuracy is impacted negatively if the white balance is offset prior to beginning a scan. This may be because the iPad has auto-adjusted the white balance to a range based on its current lighting conditions, not reflective of the scanning environment. For example, if the iPad camera was in a dark setting, and then an attempt to scan is started immediately in brighter lighting conditions without any pause for auto-adjustment, white balance inaccuracies can occur.
  
  To avoid this:
  • Before you press Start Scan, hold the iPad higher than the normal scanning distance. Wait 3-5 seconds, and if possible look at the iPad's screen to ensure the color of the fruit appears correct (as it would in real life).
  • Once confirmed, press Start Scan and then move in the regular scan motion.

Avoid initiating a scan when too close to the objects, as this tends to throw off the white balance calibration. On the screen it will look black as opposed to the desired green mesh overlay – black coverage happens when the distance is too short between the iPad and the location of the fruit.

• Focus on the centre of the bin. Avoiding the edges of the bin within the view of your scan improves the accuracy of capturing your fruit sample and shortens scan processing times.
  

• Avoid over-scanning. Continuous movement is important for a thorough scan result. We recommend moving the iPad in a circular motion over the bin, so that you have covered over the majority of the fruit within 5-10 seconds.
  
  Do not spend too much time scanning over the same spot repeatedly, as it may degrade the overall result, and well as produce large scan files which take additional processing time.

• Ensure a minimum 50-centimeter (1.5-foot) iPad distance from the bin
  you are scanning. The HQV camera has a minimum operating range;
  if you operate closer than the minimum, the pre-scan image displayed will
  show dark areas. This is an indicator that you are too close, and tracking errors may cause significant distortion issues during scanning.  Similarly, moving too far away from the bin you are scanning may capture unwanted objects outside of the bin, or degrade scan quality resolution.

• Aim for consistent lighting during your scan. When a scan is taking place outdoors, environmental conditions will vary. This is not an issue if the light remains consistent in the moment you are performing your scan. Take note of your light meter – if there is a notable change in lighting while the scan is active, the tracking process may be negatively impacted. Check if there is anything blocking the light meter (like a finger!) affecting the sensor.
  
  

• Create shade for your scan when capturing in the sun. This can mean blocking sunlight and creating shadow with one’s own body, or utilizing a shade screen or covered area.  Suitable lighting will allow the green mesh visualization over the fruit to appear easily, which means the scan is working as expected (see left image). Keeping the camera pointed at the same spot while creating shade will fill in the missing areas as they become shadowed.
  
  Poor lighting (see result example image on right) will show the fruit with unclear, black shading in the pre-scan image preview. This will make it difficult for it to produce the green mesh while scanning. 
  

• Check your scan for capture quality. You can assess the quality of your
  scan’s data capture after you tap the Stop button. Review the
  underlying shape and consistency of the scan by tapping the X-ray view
  at the bottom left of the preview screen. A failed scan in the X-ray View will
  appear rough and inconsistent (see left image).
  
  A scan that is considered to be “good” (see right image) will show
  objects as smooth and consistent. This is because the wireframe produced
  from the scan needs strong and consistent gradients to be successful.
  A “bad” scan (see right image) only has noisy, inconsistent gradients
  to compose the image, which creates the rough irregular shapes.