The question of which camera and which sensor to use should not be made based on the purchase price alone. The longer the product lifecycle should be, the more reason there is to get an industrial camera.

Cameras and sensors are available in monochrome (also called “grayscale”) and color variants. The color sensor is more expensive than a monochrome sensor with the same resolution, so it’s worth asking whether color information is really necessary in your application.

The choice of shutter method and frame rate influences the sharpness of images of moving objects. But even for applications such as counting tasks, collision prevention or aerial shots, image transmission and processing speed are more important than the perfect sharpness.

The data volume per second is determined by the image resolution and recording speed. However, it’s important to remember that this data speed is always lower at the camera’s output than the pure sensor speed!

Similar to computer chips, image sensors become more powerful with each generation. As such, it is worth studying the latest sensor technologies in the design phase and/or asking experts for advice on how sensor development is progressing.

The electrical power consumption of all system components causes heat to be generated, which needs to be taken into consideration especially for compact designs. In most cases, cooling fins or thermal bridges to the external parts of the housing are sufficient to ensure the dissipation of heat. If such measures are not feasible due to the design, then you should consider operating time and possible power-save modes.

Embedded vision is made up of four different technical components that must be attuned to one another. With the right partner companies at your side, this complexity is very easy to manage.

Of course, an image sensor or a camera is needed for an embedded vision solution. But you also need an embedded board, a carrier board, custom hardware if necessary and drivers for communication between the components – important parts that you should learn about.

There is currently no standard for MIPI CSI-2 drivers. Programming your own driver software is only recommended if your team includes experts in that area. It’s much easier to use program codes that are available in open-source communities and adapt them to your needs if necessary. The open-source CSI-2 drivers for the Allied Vision Alvium camera series are available on GitHub.

Each sensor type has a specific behavior and timing, which is why each sensor also needs a specific driver. With the Alvium camera series from Allied Vision, one driver works for all of the sensors in our product range! This is a major advantage if, for example, different sensors will be offered in the final product, or if different product variants will be implemented on a shared technological basis, such as in plant manufacturing.

Because every electronic component in the image chain also has its own processing speed, the data volumes produced are different. As such, it is important to consider data volume and processing speed throughout the chain. If the differences are too high, there can be “congestion”, or in other words, individual frames will be lost. An image buffer can help with this.

A camera offers image pre-processing, SDK (software development kit), APIs for programming plus drivers and relevant documentation. This makes the integration of a camera much simpler and faster than with a sensor module. If a fast market launch is crucial to the success of your product, there are strong arguments in favor of integrating a complete camera!

As illustrated above, cameras have a wider scope of functions than sensor modules. This enables faster integration, but the costs per unit are higher (BoM: Bill of Material). It is therefore advisable to evaluate the higher project costs for a sensor module against the higher per-unit costs of a camera in a break-even analysis. In our experience, the break-even point is often in the area of about 10,000 units of production volume.