Orientation-selective edge detection and enhancement using the irradiance transport equation

Polarization-based spatial filtering for directional and nondirectional edge enhancement using an S-waveplate

Using polarization as an additional parameter apart from amplitude and phase in spatial filtering experiments offers additional advantages and possibilities. An S-waveplate that can convert a linearly polarized light into radially or azimuthally polarized light can also be used for isotropic edge enhancement. For anisotropic edge enhancement, introduction of a polarizer at the output was recommended and edge selection was done by orientation of the polarizer. But the full potential of the S-waveplate as a spatial filter has not been exploited so far. Unlike the standard amplitude and phase-based Fourier filters, which are independent to the state of polarization of the illuminating beam, the S-waveplate acts in a different way depending on the state of polarization. The edge selection does not need to be carried out by changing the orientation of the polarizer. With a fixed polarizer at the output, we show that either isotropic or anisotropic edge enhancement in any desired orientation can be performed by operating the same spatial filter setup in different illuminating polarization states.

Fractional spiral zone plates

In this paper, we generalize the concept of classical spiral zone plates (SZPs) to fractional spiral zone plates (FSZPs). By using an SZP with a fractional topological charge and controlling the starting orientation, we can break down the symmetry of the focusing process to give orientation-selective anisotropic vortex foci. Numerical results show that its binary structure gives additional high-order foci on the optical axis and the intensities in the foci can be controlled by properly choosing the fractional topological charge. Our study reveals the feasibility to control the intensity in the foci by means of FSZPs.

Edge enhancement of color images using a digital micromirror device

A method for orientation-selective enhancement of edges in color images is proposed. The method utilizes the capacity of digital micromirror devices to generate a positive and a negative color replica of the image used as input. When both images are slightly displaced and imagined together, one obtains an image with enhanced edges. The proposed technique does not require a coherent light source or precise alignment. The proposed method could be potentially useful for processing large image sequences in real time. Validation experiments are presented.

Incoherent optical processor for nondirectional edge enhancement of color images

We present an optical method for nondirectional edge extraction/enhancement in color images. The method is based on the capability of twisted-nematic LCDs to traduce the image information in changes of the state of polarization of light, which allows us to generate simultaneously two replicas of the digital image displayed on the LCD: a true-color ("positive") image and a complementary-color ("negative") one. In our setup the imaging system consists of a lens plus a pupil mask formed with concentric apertures and orthogonal polarizers. This layout allows us to simultaneously image a well-focused positive replica (due to the circular aperture) superimposed to a slightly defocused negative one (due to the annular aperture). It is not difficult to demonstrate that this generates a nondirectional (Laplacian) edge enhancement. Unlike Fourier, our proposal works with incoherent illumination and does not require precise alignment, and thus, it could be a useful tool for edge extraction/enhancement in large images in real-time applications. Validation experiments are presented.

Optical processing of color images with incoherent illumination: orientation-selective edge enhancement using a modified liquid-crystal display

We present a novel optical method for edge enhancement in color images based on the polarization properties of liquid-crystal displays (LCD). In principle, a LCD generates simultaneously two color-complementary, orthogonally polarized replicas of the digital image used as input. The currently viewed image in standard LCD monitors and cell phone's screens -which we will refer as the "positive image or true-color image"- is the one obtained by placing an analyzer in front of the LCD, in cross configuration to the back polarizer of the display. The orthogonally polarized replica of this image -the "negative image or complementary-color image"- is absorbed by the front polarizer. In order to generate the positive and negative replica with a slight displacement between them, we used a LCD monitor whose analyzer (originally a linear polarizer) was replaced by a calcite crystal acting as beam displacer. When both images are superimposed laterally displaced across the image plane, one obtains an image with enhanced first-order derivatives along a specific direction. The proposed technique works under incoherent illumination and does not require precise alignment, and thus, it could be potentially useful for processing large color images in real-time applications. Validation experiments are presented.

Nondirectional edge enhancement by contrast-reverted low-pass Fourier filtering

We present an image processing method for nondirectional edge extraction/enhancement. The method is based on the capability of twisted-nematic liquid-crystal displays (LCDs) to traduce the image information in changes of the state of polarization of the light, which allows us to generate simultaneously a "positive" and a "negative" (i.e., contrast-reversed) replica of the digital image displayed on the LCD. The negative image is low-pass filtered in a novel polarization-selective 4f optical processor. When the smoothed negative image is imagined together with the original image, an image with nondirectional edge enhancement is obtained. Unlike other Fourier methods presented in the literature, the proposed technique provides a simple way to control the relative amount of high frequencies present in the final image. The proposed method does not involve numerical processing, and, thus, it could be a useful tool for edge extraction/enhancement in large images in real-time applications. Validation experiments are presented.