Spatial Frequency Enhancement Research Articles

Spatial frequency enhanced salient object detection

Many salient object detection (SOD) methods based on convolutional neural networks utilize spatial frequency information from images to obtain salient features. However, most of these methods do not extract and/or exploit spatial frequency information effectively. This results in suboptimal detection performance in complex natural and human social scenes. In this work, we first propose an effective and flexible spatial frequency enhancement (SFE) module based on generalized Oct-convolution. SFE can effectively extract and incorporate multiple spatial frequency information from in-stage and cross-stage feature maps and output an arbitrary number of comprehensive and compact frequency features. Then we design a spatial frequency enhanced network (SFENet) adopting two SFE modules to refine high- and low-frequency salient features end-to-end. High-and low-frequency salient features are then integrated into a final full-band saliency prediction. Experiments show that SFENet can ensure the integrity and sharpness of salient regions in complex backgrounds on five benchmark datasets.

Image Focus Measure by Wavelet Transform with Gaussian Derivative and Spatial Frequency and Image Enhancement using Fuzzy Logic

Image Focus Measure by Wavelet Transform with Gaussian Derivative and Spatial Frequency and Image Enhancement using Fuzzy Logic

Selective modulation of visual sensitivity during fixation.

During periods of steady fixation, we make small-amplitude ocular movements, termed microsaccades, at a rate of 1–2 every second. Early studies provided evidence that visual sensitivity is reduced during microsaccades—akin to the well-established suppression associated with larger saccades. However, the results of more recent work suggest that microsaccades may alter retinal input in a manner that enhances visual sensitivity to some stimuli. Here we parametrically varied the spatial frequency of a stimulus during a detection task and tracked contrast sensitivity as a function of time relative to microsaccades. Our data reveal two distinct modulations of sensitivity: suppression during the eye movement itself and facilitation after the eye has stopped moving. The magnitude of suppression and facilitation of visual sensitivity is related to the spatial content of the stimulus: suppression is greatest for low spatial frequencies, while sensitivity is enhanced most for stimuli of 1–2 cycles/°, spatial frequencies at which we are already most sensitive in the absence of eye movements. We present a model in which the tuning of suppression and facilitation is explained by delayed lateral inhibition between spatial frequency channels. Our data show that eye movements actively modulate visual sensitivity even during fixation: the detectability of images at different spatial scales can be increased or decreased depending on when the image occurs relative to a microsaccade.NEW & NOTEWORTHY Given the frequency with which we make microsaccades during periods of fixation, it is vital that we understand how they affect visual processing. We demonstrate two selective modulations of contrast sensitivity that are time-locked to the occurrence of a microsaccade: suppression of low spatial frequencies during each eye movement and enhancement of higher spatial frequencies after the eye has stopped moving. These complementary changes may arise naturally because of sluggish gain control between spatial channels.

Active Vision: Dynamic Reformatting of Visual Information by the Saccade-Drift Cycle

Visual processing depends on rapid parsing of global features followed by analysis of fine detail. A new study suggests that this transformation is enabled by a cycle of saccades and fixational drifts, which reformat visual input to match the spatiotemporal sensitivity of fast and slow neuronal pathways.

Experimental opto-digital synthesis of encrypted sub-samples of an image to improve its decoded quality

We develop an experimental protocol to visualize decrypted images that otherwise would have been barely recognizable, while keeping the standard security levels. This image deterioration is partly due to the natural speckle noise as well as the practical limitations arising from the optical elements composing the setup. This protocol is based on an optical image synthesis with digital holography using enlarged sub-samples of an entire image together with a multiplexing technique. We implement the process using a Mach-Zehnder interferometer with a joint transform correlator encrypting architecture. As a result, we get smaller speckle patterns on the final assembled image and a spatial frequency enhancement with respect to the decoded image obtained with the conventional procedures.

Spatial frequency processing conditions on intraoral computed radiography

Radiographic information on the imaging plate was read with a pixel size of 50 μm×50 μm and recorded with 100 μm×100 μm. Various frequency-enhanced images were produced by an unsharp masking process that can maximize the spatial frequency enhancement at 0.25, 0.50, 1.00, 2.00, and 4.00 cycles/mm. Using these images, the effect of frequency enhancement processing for the X-ray diagnosis of normal anatomical structures and simulated bone changes on intraoral computed radiography was investigated. The imaging conditions under which the most valuable radiographic images for diagnosis of normal anatomical structures were obtained was a frequency type that diminished the low level density area, a non-linear gradation and an enhancement rank of 4 at the spatial frequency of 2.00 cycles/mm. These iamges projected a higher image quality than conventional periapical films. In addition, the intraoral computed radiography had equivalent or better ability to detect fine-simulated bone changes compared to conventional periapical films. These findings suggest that the spatial frequency enhancement area needed to obtain images with significant diagnostic value, equivalent to or better than conventional periapical film images, is 2.00 cycles/mm in intraoral computed radiography, with maximal spatial resolution of 5.00 cycles/mm.

Digital Luminescence Radiography and Conventional Radiography in Abdominal Contrast Examinations

In 326 patients abdominal contrast radiographs were compared to digital luminescence radiographs (DLR) and conventional screen-film system ones. The digital exposure dose was 50% of the conventional. In DLR, 2 different types of postprocessed images were obtained from each data set. A display with low spatial frequency enhancement filtered to look like a conventional radiograph was compared to a display with high spatial frequency enhancement. Conventional and DLR images were evaluated randomly and separately by 4 radiologists by means of a questionnaire. DLR proved to be diagnostically equivalent to the conventional technique with the exception of a slightly diminished visibility of the mucosal pattern. High spatial frequency enhancement did not provide additional diagnostic information and should be dispensed with in abdominal examinations.

Computed radiography: User-programmable Features and Capabilities

Digital or computed radiography (CR) using photostimulable storage phosphor plate technology is becoming increasingly popular in certain clinical applications, such as bedside radiography, where it possesses clear advantages over conventional screen-film imaging. The majority of CR systems in clinical use have been manufactured by Fuji Medical Systems USA, Inc (Stamford, CT) and provide a surprising degree of flexibility. Fuji CR units are delivered with preset menus, hardcopy format, and image-processing parameters for each examination. Of practical importance is that users may change the exam menu and printed film format as well as the image-processing parameters for each examination. There is, however, a lack of documentation describing these features and how they are programmed. This paper addresses these issues. Examples are given on how to change: 1) the printed film format, 2) the contrast and gray-scale processing, 3) spatial frequency enhancement, and 4) the appearance of the operator interface menus.

Röntgenuntersuchungen der peripheren Gelenke: Vergleich der digitalen Lumineszenzradiographie (DLR) mit Filmfoliensystem

175 matched skeletal x-ray examinations were carried out by digital luminescent radiography (DLR) and conventional screen-film system combinations. In digital luminescent radiography (DLR) two differently postprocessed images were obtained from one x-ray exposure: A display with low spatial frequency enhancement was opposed to a second display with high edge enhancement. Conventional and digital images were evaluated randomly and separately by four radiologists using a questionnaire. DLR proved to be diagnostically equivalent to the conventional technique except for a diminished visibility of subtle erosions and fissures. As major advantages of DLR maybe pointed out that faulty exposures were avoided by the automatic adjustment of image brightness and that the soft tissue margins could be more easily delineated in displays with high edge enhancement than in conventional radiography. However, digital displays with high edge enhancement did not provide additional diagnostic information and led to artifacts at the edges of metal prostheses.

Mediastinal abnormalities: detection with storage phosphor digital radiography.

Conventional film radiography (FR) and six postprocessing algorithms of isodose storage phosphor digital radiography (SR) (0.2-mm X 10-bit pixel matrix) were compared in the evaluation of 40 mediastinal and 30 pulmonary lesions in 60 patients who underwent computed tomography of the chest. The six SR algorithms varied among each other in only one image parameter. One algorithm approximated conventional image characteristics. The other five algorithms were designed to optimize imaging of the mediastinum and tested the effects of gray-scale reversal, adjustment of optical density, a linear instead of a sigmoid gradation curve, and moderate edge enhancement of high and medium spatial frequencies. Performance was evaluated by calculating the average area under the receiver operating characteristic curve (Az) of 5,040 observations by six readers. Post-processing with high-frequency edge enhancement and density optimization for the mediastinum significantly improved performance of SR over FR in the detection of mediastinal lesions (Az = .80 +/- .02 vs .73 +/- .01, respectively). Gray-scale reversal significantly decreased performance (Az = .64 +/- .03). All SR algorithms that were postprocessed to optimize imaging of the mediastinum were significantly inferior to FR in the detection of pulmonary lesions.

Kolonkontrasteinlauf in digitaler Lumineszenzradiographie (DLR) und konventioneller Röntgentechnik

120 double-contrast barium enemas were obtained by both digital luminescent radiography (DLR) and conventional screen-film systems, the digital exposure dose being 50% of the conventional one. In DLR two differently post processed images were obtained from one x-ray exposure: a display with low spatial frequency enhancement was processed to look like a conventional radiograph and was complemented by a display with high spatial frequency enhancement. Analysing the results statistically DLR proved to be diagnostically equivalent to conventional radiography despite the reduction in exposure dose and a slightly diminished image quality. High spatial frequency enhancement did not provide further diagnostic information and is therefore superfluous in barium enemas.

Noncoherent Talbot effect: Coherence theory and applications

We discuss the coherence theory of the bands generated by using an incoherently illuminated grating at infinity, and other gratings at finite distances. Our discussion is applied to the design of a noncoherent technique for spatial frequency enhancement, and a novel interferometer. Experimental results are included.

Digitale Radiographie in der Thoraxdiagnostik

Technologies and performance of digital projection radiography are reviewed. Storage phosphor imaging plates can be used in existing diagnostic equipment and the resulting digital images permit the recognition of fine interstitial changes, if a high frequency edge enhancement is performed. While a generalized enhancement of low and medium spatial frequencies and a grey scale reversal do not improve the detectability of pulmonary nodules, selective enhancement of low frequencies in the mediastinal and retro-diaphragmatic areas may well improve diagnostic performance. Dose can be reduced if the resulting signal-to-noise ratio is adequate for diagnosis. Further dose savings can be achieved by the reduced number of retakes. Dual energy techniques improve the detectability and differential diagnosis of pulmonary nodules. The further development of picture archiving and communication systems (PACS) will turn the digital projection radiography into the method of choice for the diagnosis of chest disease.

Optimization of readout rate of television cameras in pulsed digital radiography

Adaptation of existing fluoroscopic television rooms to permit radiographs to be recorded directly by the television system would be a very practical way to introduce digital radiography into general use if the image quality was adequate. The main limitation of current systems is spatial resolution. Although a digital system can change the spatial frequency response by application of spatial frequency enhancement, the amount of enhancement which can be usefully applied is a function both of the spatial frequency response of the whole system and image noise. Slow scan television systems have been proposed as a way to reduce noise; and hence, as a way to potentially improve limiting resolution. We examine all the consequences of changing readout scan rate, both theoretically and with measurements on a commercial camera with a lead oxide image tube. It is found that recent improvements in camera tube design, specifically the increase in current capability of the electron gun, have reduced the advantages theoretically available with slow scan. However, an order of magnitude increase in useful dynamic range is still possible.

Ｄｉｇｉｔａｌ ｒａｄｉｏｇｒａｐｈｙによる喉頭の画像診断

The author evaluated the usefulness of digital radiography in the radiographic diagnosis of laryngeal diseases, utilizing Fuji Computed Radiography (CR) system. Main distinctive features of CR are as follows; photostimulable phosphor is used as the memory material, X-ray energy pattern is converted into digital signals, and then image signals are processed digitally.As the result, three types of image processing of CR, i. e. contrast processing, spatial frequency enhancement, and image subtraction, were clinically useful methods. Especially, one shot dual energy subtraction method is a new technical procedure to extinguish both cervical spine and calcified cartilage shadows in plain antero-posterior laryngeal X-photograph. Although this method belongs to a kind of plain X-photography, the feature of the laryngeal soft structures can be clearly demonstrated without little mortion artifact during examination.Digital radiography has so high spatial resolution and various image processing systems that it is possible to replace this system for some of the conventional screen-film method in X-ray analysis of laryngeal diseases.

Optical Processing of Lattice Images

An optical image processing facility designed specifically for the treatment of lattice images is described. A range of optical processing techniques for improving the interpretation of lattice images are outlined and illustrated with examples. These operations include methods of signal-to-noise enhancement and pseudocolour encoding of spatial frequencies.

A scanning system for chest radiography with regional exposure control: theoretical considerations.

Conventional chest radiography is limited by the presence of scattered radiation and the small useful exposure range of radiographic film. A computer-assisted scanning system to minimize these two effects is outlined. The system uses a small beam of radiation swept over the patient's chest in a raster pattern to expose a conventional film cassette, while a slit collimator scanning between the patient and the film serves to reject scattered photons. A microcomputer measures beam attenuation by the patient with a detector placed behind the film which in turn automatically adjusts the x-ray tube output to minimize excursions in film exposure as the beam scans. A formalism which relates the patient transmission and film exposure distribution is developed and a system transfer function is given. It is shown that such a system operates as a spatial filter which attenuates film contrast for structures of spatial frequency less than the inverse scanning beam width. By manipulating the software parameters of the feedback network, it is possible to alter this filter and produce radiographs with low spatial frequency enhancement, attenuation, or contrast inversion.