In a compact digital lensless inline holographic microscope (LIHM), where the sample-to-sensor distance is short, the imaging resolution is often limited by sensor pixel size instead of the system numerical aperture. We propose to solve this problem by applying data interpolation with an iterative holographic reconstruction method while using grating illumination in the LIHM system. In the system setup, the Talbot self-image of a Ronchi grating was used to illuminate the sample, and the inline hologram was recorded by a CMOS imaging sensor located behind the sample. The hologram was then upsampled by data interpolation before the reconstruction process. In the iterative holographic reconstruction, the sample support was defined by the bright areas of the grating illumination pattern and was used as constraint. A wide-field image can also be obtained by shifting the grating illumination pattern. Furthermore, we assumed that the sample was amplitude object, i.e., no obvious phase change was caused by the sample, which provided additional constraint to refine the interpolated data values. Besides improved resolution, the iterative holographic reconstruction also helped to reduce the twin-image background. We demonstrated the effectiveness of our method with simulation and imaging experiment by using the USAF target and polystyrene microspheres with 1 μm diameter as the sample.
High resolution is always a pursuing target in the imaging field, as a new prospective technique in imaging applications, digital in-line holography has become a very active field for compactness, more information and low-cost. However, for compact system, the resolution is often limited by sensor pixel size. To overcome this problem, we propose an iterative reconstruction method with data interpolation based on the grating illumination. In our method, the Talbot self-image of a Ronchi grating is exerted in the sample plane as a priori constraint which lead to the convergence of the iteration, the iteration between the sample plane and the sensor plane can provide some extra information with interpolation in the sensor plane based on the a priori constraint, furthermore, the iteration reconstruction can also eliminates the twin-image to improve the image quality. Numerical simulation has been conducted to show the effectiveness of this method. In order to make a further verification, we have developed a lensless in-line holographic microscope with a compact and wide field-of-view design. In our setup, the sample was under the Talbot image illumination of the Ronchi grating, which was illuminated by a collimated laser beam, and holograms were recorded by a digital imaging sensor. We can shift the grating laterally to get a wide-field image. We demonstrated the resolution of our imaging system by using the USAF resolution target as a sample, and the results shown the resolution improvement of the image.
We report a new holographic microscope using pixel super-resolution algorithm. In our method, a sequence of low resolution images are acquired by a complementary metal oxide semiconductor (CMOS) sensor in digital inline holography system and the resolution is limited by the sensor pixel size. Then the super-resolution algorithm is applied to the low resolution images to get the image with much higher resolution that beyond the Nyquist criteria. We perform both numerical simulation and experiments to demonstrate our method with US Air Force Target used as the sample. The sample is randomly moved in the sample plane and a set of holograms are captured by the camera in inline holographic system. We use two methods to reconstruct the sample image. In the first method, super-resolution algorithm is applied with the low resolution holograms to get the high resolution hologram. Then the high resolution hologram is reconstructed using auto-focusing algorithm to get the high resolution sample image. In the second method, the raw holograms are directly reconstructed to get a set of low resolution sample images, then the super-resolution algorithm is applied to get the high resolution sample image. We observed that the above mentioned two methods can get similar results in both numerical stimulation and experiments. We believe that the combination of pixel super-resolution algorithm and digital in-line holography can be very useful to implement a compact low-cost microscope with high resolution.
Digital in-line holography (DIH) is a lensless imaging technique that can be used to build low-cost and compact imaging systems. In DIH, the in-line hologram is recorded by a CMOS or CCD sensor and later used to reconstruct the image of the sample. The imaging resolution is determined by the system numerical aperture provided that the pixel size is smaller than the required Nyquist criteria for sampling distance. In the case of short sample-to-sensor distance, pixel size is often a limiting factor for the resolution. To solve this problem, we propose to use iterative method along with data interpolation for the holographic reconstruction. Proof-of-concept numerical simulations have been done to show the effectiveness of our method. In our algorithm, the optical field is propagated back and forth between the sample plane and the sensor plane while using the measured intensity and a priori information about the sample as constraints, following Gerchberg-Saxton and Fienup’s methods. The iteration will converge and we can get both intensity and phase information of the sample. Before the iteration, the intensity data matrix measured by the sensor is interpolated to enlarge the matrix dimension and thus effectively reduce the pixel size. During the iteration, we apply the sensor plane constraints on only the measured intensity location but not the interpolated data location. In our simulation, we observed that during the iteration, the interpolated data will be changed reasonably and we can finally reconstruct the sample image with better resolution.
We propose a novel large-scale conditional random field model with respect to the problem of natural outdoor scene labeling. The novelty of the proposed method lies in three aspects: 1. features from two neighboring regions are concatenated to form the input of the pair-wise classifier to compensate for the simultaneous feature deviation of neighboring regions; 2. the definition of a generalized neighboring system and the incorporation of direction-specific patterns in conditional random field models based on the generalized neighboring system to better simulate the visual cognition of human being; and 3. the definition of a similarity criterion based on the bags-of-words expression to facilitate the incorporation of semantic patterns. The proposed model is first evaluated over the Corel dataset. Both qualitative and quantitative results show that our model is capable of modeling large-scale spatial relationships between objects in natural outdoor scenes, and achieves better results than other existing conditional random field models. Furthermore, our model is also evaluated over several other natural datasets, which are taken from logged field tests, to further demonstrate the adaptability of our model to different lighting conditions.