Introduction

10 Best Practices for Image Capture

Here are a few considerations for taking images to ensure we can build a pipeline to extract the highest quality and quantity of data from images.

1. Maintain a consistent angle, distance, and field of view.

To maximize the robustness of your subsequent analysis pipeline, it is crucial to maintain a consistent angle, distance, and field of view for all images in a set. To achieve this, mounting the camera at fixed distance from the object of interest is ideal. Taking images of objects at an angle can create a situation where features in the foreground are at a different scale than features in the background of the field of view. To maintain consistent scaling throughout the field of view, position the camera so the focal plane is parallel to the object of interest.

Open

Figure 1. An adjustable camera stand is a versatile piece of equipment for consistent imaging. A camera stand (79 cm width × 82.5 cm height) can be built from aluminum framing (80/20 Inc., Columbia City, Indiana, USA) to hold a DSLR camera via a standard mount (Adapted from Tovar et al., 2018).

Open

Table 1. 80/20 Inc. (https://8020.net Columbia City, Indiana, USA) Parts list to build camera frame shown in Figure 1 (adapted from Tovar et al., 2018).

2. Avoid handling the camera.

Employ a remote shutter release or camera control software via computer to avoid handling the camera once it is set up. This will prevent accidental shifts in field of view, focus, or internal camera settings across the image set.

3. Use consistent camera settings.

Use consistent camera settings across all images. Avoid using “Auto Mode” on your camera as the shutter speed, aperture, ISO, white balance, and flash settings may change from image to image. Set your camera up to ensure you are using the same settings across all your images (aperture, ISO, shutter speed, white balance, magnification, focus).

4. Use a photo booth.

To prevent shadows and inconsistent lighting across your batch of images, consider using a “photo booth” with controlled lighting. This is especially important if taking images in an outdoor setting where angles and intensity of lighting are subject to change over time.

5. Use a contrasting background.

Utilize a solid-colored, non-reflective (matte) background that contrasts in color from the object you are photographing (White, Black, or Blue can be good choices). Matte-white is often a good choice for a background but in circumstances where this may be subject to dirt, blue or black may be a better choice.

6. Include a size marker.

Include a size marker in the same location of each image for scale calibration (ideally a corner of the field of view. This size marker can be used to normalize object area during image analysis.  We recommend using adhesive Tough-Spots (1.27 cm/0.5 in diameter) for a scale marker (Research Products International, Mount Prospect, Illinois, USA). These recommended scale markers can be purchased here:

https://www.rpicorp.com/products/laboratory-equipment/labels/tough-spots-1-2-dia-green-1000-pk.html

7. Include a color card.

If color is to be analyzed, include a color card in a consistent location within the field of view for color correction (white, gray, and black; DGK Color Tools Optek Premium Reference White Balance Card; DGK Color Tools, New York, New York, USA). This color card affords the ability to normalize white balance in situations where lighting conditions may change across your batch of images.

Recommended color card can be purchased here: https://www.bhphotovideo.com/c/product/1014565-REG/dgk_color_tools_wdkk_bubble_bag_1.html/?ap=y&ap=y&smp=y&smp=y&lsft=BI%3A514&gclid=Cj0KCQiAosmPBhCPARIsAHOen-Mg5l6pSjZO7ckPq2A2DeFQcRE-VMjFbSda9brPZ74b1MMvOly2GI4aAv20EALw_wcB

8. Mark the region of interest.

Defining and marking a “region of interest” or location where objects will be placed on this background helps maintain consistent positioning of objects within the field of view across a batch of images, thereby making the process of identifying objects in the subsequent analysis pipeline easier.

9. Identify samples with both human-readable information and bar codes.

Consider including some human readable information to identify the sample in each image (sample tag or label). Barcodes and QR codes may also be included and software to read these in images is currently in development. Including human readable “tags” in each image can serve as a fail-safe to verify the identity of each sample for future QC or use in presentations or figures. Similar to the scale marker and color card, place labels with identifying information in a consistent location within the field of view from image to image.

10. Separate objects.

If multiple objects will be included within a single image, space the objects out across the background so they do not touch or overlap with one another. Separating objects will greatly increase the efficiency of segmentation in the analysis pipeline and allows each object within the image to be treated as a distinct data point (each object is given equal representation within an image allowing for determination of variation within each object as well as across all objects in an image).

Related articles