Hitachi Develops Lens - Less, Ultra - Slim Camera


Hitachi Develops Lens-less, Ultra-slim CameraHitachi Ltd developed a new image-sensing technology to reduce the size and cost of cameras for its "social innovation" business.
With the new technology, the company aims to take images in more locations, collect them as data, analyze them by using AI (artificial intelligence), etc and use them as valuable data.


Effective in automotive, monitoring applications

The new technology eliminates the need for an optical lens by combining an existing image sensor and image data processing. It uses a thin film with a thickness of, for example, several tens of micromillimeters in place of a lens. There is a concentric pattern on the film, and the pattern and image data processing function as a lens.

The technology is expected to reduce the size and cost of cameras by eliminating a lens, which accounts for most of the size and cost of a camera. Especially, it is useful in making a large image sensor capable of taking images in dark environments for monitoring cameras or automotive applications because it enables to replace multiple (thick) large-diameter lenses with a thin film.

The distance between the surface of an image sensor and the film is about 1mm, making it possible to drastically reduce thickness, compared with a case where a lens is used.

It is also possible to focus on a desired point after taking an image. This is possible because of the processing method used to reproduce images from data that the image sensor collects via the film.

In general, the method that Hitachi employed for its lens-less camera uses a "moire stripe" that can be obtained by stacking two concentric-patterned films with a certain interval and transmitting light through them. The numerous light-emitting points constituting the image influence the pitch and orientation of the moire stripe. The location of light, etc can be restored by applying two-dimensional Fourier transformation to the moire stripe.

This time, Hitachi replaced one of the films (one that is closer to the image sensor) with image processing. In other words, one film is placed with an interval of about 1mm, but the other film does not actually exist. And, instead of using the second film, a concentric pattern is superimposed on image data.

To reproduce an image from the moire stripe through Fourier transformation, it is necessary to match the diameters of the two concentric patterns on the surface of an image sensor. And this changes depending on the distance from an object. Moreover, if they coincide, the moire stripe becomes straight. If not, it is curved.

With the new technology, a point where a moire stripe becomes straight is found by changing the diameter of the concentric pattern through image processing. And this is the state of being in focus. A point can be focused by straightening a moire stripe created by light in the point.

The latest method does not require high processing capability because the established high-speed processing method (two-dimensional Fourier transformation) is used. Compared with a case where a film with a pattern other than a concentric pattern is used, the required processing capability is 1/300, Hitachi said. Also, the processing capability for displaying video in real time can be provided by a microprocessor of a personal computer or embedded microcomputer, the company said.