Search

A sparse representation-based local occlusion recognition method for athlete expressions is proposed to address the problems of large mean square error, low recall rate, and poor recognition performance. We calculate the gradient direction and size of image pixels, divide image blocks, count the histogram of the gradient direction of the image blocks, combine all small histograms into a feature vector, and obtain the facial feature extraction results. The LBP algorithm is used for local occlusion image segmentation, and a sparse representation model is established to extract expression features. By dividing the image into blocks and solving the sparse representation coefficients of each block, local occlusion expression recognition is achieved. Experimental results show that the maximum mean squared error of the proposed method for facial expression recognition is only 0.21, and the maximum recall rate is more than 80%, which shows that it can effectively recognise occluded parts.

Though surface plasmon resonances can reduce image beam width of a plasmon superlens, they distort the transfer function in the spectrum and introduce large image sidelobes as well. In this paper, image enhancement of a plasmon film superlens is investigated by adding permittivity loss. First, we add the loss in the film of the superlens and observe the sidelobes suppression. Second, we introduce loss in the image region of the superlens device and observe a flatter transfer function and obtain improved image resolution. For the silver film superlens at a free space wavelength of 337.5 nm, a beam width reduction of 69% is observed. Previously, we found that introducing roughness in the superlens can reduce the beam width. In this paper, we combine surface roughness with the method of adding loss in the image region and observe a further beam width reduction. The lossy sinusoidal surface superlens at a wavelength of 351 nm gains a beam width reduction of 86% compared to the lossless flat superlens. Moreover, in this paper we provide a model for calculating the superlens near-field image intensity when the objects are illuminated by a laser source, and the more general five slits example is shown to further demonstrate the advantage of adding loss.

The loss and back-coupling effects on the subwavelength imaging of three-dimensional superlens are reported in this paper. The loss is added in the image region of a superlens. The back-coupling effects are considered by adding a shielded layer above the object region. (1) By adding loss in the image region, the long range plasmon mode is drastically suppressed. (2) The back-coupling shield above the objects has the effects of amplifying the higher spatial frequency components while suppressing the long range plasmon mode. Because of (1) and (2), the transfer function becomes flatter. Subsequently, the finer resolution of images is obtained. This is confirmed by the field and intensity distribution generated by the horizontal magnetic dipoles and vertical electric dipoles located in the object region and the image intensity distributions of the patterned mask structures in the lithography.

In this Letter we investigate the subwavelength imaging of a three-dimensional plasmon superlens based on the full vector wave simulations of optical wave propagation and transmission. The optical transfer functions are computed. Comparisons are made between the results of lenses with flat and periodic/random rough surfaces. We also study the problem of practical imaging system geometry using laser as an illumination source. Results show that the lens with periodic or random roughness can reduce the field interference effects, and provide improved focus on the transmission field and the Poynting flux. We illustrate that the subwavelength roughness in a plasmon lens can enhance the image resolution over a flat lens for both matched and unmatched permittivity conditions. The enhancement of resolution occurs because the introduced subwavelength roughness can amplify the evanescent wave components and suppress the surface plasmon resonance peaks.