A high resolution image containing spatial and spectral information (wavelength and intensity of the radiation) is known as data cube. Data cube has become a powerful tool in almost every science and technology field. If a data cube is generated in time sequence as in scanning spectrometer, besides being at disadvantage for having mechanically moving parts or moving platform with respect to object, the exposure is reduced and different frame at different time may not provide true information. Snapshot type Imaging Spectrometer (SIS) systems that include Snapshot Hyperspectral Imaging (SHI) systems and Snapshot Spectral Domain Optical Coherence Tomography (SSD-OCT) systems, as an example, are proving advantageous over scanning and Fourier domain type. A large volume of literature is available on this topic and many references are listed in recent U.S. Pat. No. 8,233,148 to Bodkin et al. and U.S. Pat. No. 8,174,694 to Bodkin for SHI systems and US Patent Application Number 2013/0250290 to Tkaczyk et al. for SSD-OCT systems. These documents provide description of SHI and SSD-OTC systems and their importance and applications of data cube as well as many other references.
SHI systems disclosed by Bodkin in U.S. Pat. No. 8,174,694 utilize a cylindrical lens array and/or slit array near the image from front camera. This way, each image pixel row (column) gets compressed and nearly illumination free spaces between rows (columns) are generated. After pixels are dispersed with dispersing element and reimaged on focal plane array, rows (column) fill in blank spaces according to rows (column) spectral content. In this arrangement the number of spaces available to fill blank spaces between rows (columns) is to equal or less than optical compression factor of the cylindrical lens and/or slit array. That determines the number of spectral channels possible according to Bodkin's patent. Variations of the basic scheme are discussed in Bodkin's patent to have cylindrical lens array and/or slit array at position near detector array. Bodkin also provides different means by which dispersion and formation of image bars can be accomplished. All these approaches are generally applicable for arrangements and fabrication of similar other SIS systems besides SHI systems.
Since, Bodkin in U.S. Pat. No. 8,174,694 uses cylindrical lens array and/or slit array, image pixels have very large aspect ratio and require detectors in Focal Plane Array (FPA) of similar geometry. Alternatively, one can use an additional cylindrical lens to bring pixel's aspect ratio near 1:1 and select detector array with nearly square or round detector geometry, or use higher spatial resolution in that direction. However, the FPA aspect ratio would be large in that case.
Tkaczyk et al. disclose use of an image mapper with dispersive reimager to achieve snapshot operation of spectral domain optical coherence tomography systems. According to their discloser, interfering Electro Magnetic Radiation (EMR) image pixels received from front optics are divided and reflected in groups by an image mapper (multi-faceted mirror) in various distinct directions creating nearly illumination free spaces between redirected adjacent groups when dispersion is absent. After spectral dispersion and re-imaging, dispersed image pixels fall on distinct detector of the FPA.
In Tkaczyk et al.'s case of SSD-OCT system, the image at the image mapper is formed by the preceding optics from depth encoded interfering EMR produced either as emitted, back-scattered, reflected or scattered. In Bodkin's case, the image is formed at or near the cylindrical lens array and/or slit array by the preceding optics simply from incident EMR either emitted, back-scattered, reflected or scattered from an object. As for the SIS systems the EMR at the input from either system appears similar in nature; hence, the EMR input would mean same as image formed by the input optics at the image plane or position and may be referenced as input. The process from this input to FPA is same in both cases for SIS. The image mapper or cylindrical lens array and/or slit array can work in either case with number of spectral channels equal to their specific design. Elaborating further, the image mapper in an SSD-OCT system as disclosed by Tkaczyk et al. if is replaced with a cylindrical lens array and/or slit array in SHI system as disclosed by Bodkin in U.S. Pat. No. 8,174,694 or vice versa, basic functions of both those systems would be same. Addition of a multi-faceted mirror and accompanying complexity makes SSD-OCT system as disclosed by Tkaczyk et al. is more expensive and difficult to realize.
Bodkin et al. in U.S. Pat. No. 8,233,148 disclose use of lenslet and/or aperture array in place of cylindrical lens and/or slit array that were disclosed in U.S. Pat. No. 8,174,694. Bodkin et al. mention rotating the dispersing direction (i.e., angle of dispersive element relative to focal plane) to avoid overlap of different spectra on detector. However, they fail to teach how an oblique dispersion can be used to spread spectrum over nearly N times the number of rows (or columns) to achieve˜N×N spectral channels and closer to square optical, mechanical and photo-detector form factor, where N is the compression factor of the image pixel by lenslet and/or pinhole array. From the mechanical and optical design consideration it is advantageous to have dimension of image and pixels in both the directions (X and Y) approximately equal. My invention clearly shows how to achieve this result.
A larger number of spectral channels are desirable for accurate analyses in many instances, such as predicting chemical compositions, burn signatures, forensic, biomedical and genetics. Consider visible spectrum from 450 nm to 700 nm that is a range of 250 nm. Spectral resolution of 5 nm and sometimes even 1 nm is desirable. A SHI system with pixel compression of 6, and Bodkin et al.'s illustration for spectral spread would produce roughly 12 spectral channels. Roughly 36 spectral channels are possible with my invention disclosed here.
Data cube generated from SIS systems has three dimensional (3D) aspects. Mathematically, it is a 3D matrix with matrix element value corresponding to the radiation intensity. Two spatial dimensions and wavelength dimension constitute three dimensions of the data cube. In all SIS systems, image pixels are dispersed such that each dispersed pixel within a spectral band or channel nearly fill one detector of the detector array. Digital signal processing allows us to re-format the data so that a desired scene, such as a two dimensional (2D) color image where colors correspond to wavelengths in case of SHI systems, or a three dimensional (3D) object with depths decoded from wavelengths in case of SSD-OCT systems may be displayed. As long as dispersed pixel formation on the detector array is known, re-formatting can be done for any pattern. In U.S. Pat. No. 8,174,694 each pixel line is dispersed along row (column) direction. The possible number of spectral channels is limited by the size of the image and the spectral dispersion of the said pixel until it starts to overlap other pixel's spectrum. The cylindrical lens arrays and/or slit arrays have physical limits for its size and focusing power. Therefore, more than 20 spectral channels are difficult in that case. In Tkaczyk et al.'s case of SD-OCT system the spectral interferogram image is sliced by image mapper having several sets of group of mirrors with distinct angular positions within the group. In effect, the scheme is similar to dispersing along rows (columns) direction except that aspect ratio of the image at FPA is improved by dividing long ribbon of dispersed images into manageable lengths and placing segments side by side at FPA by means of image mapper. Complexity of image mapper, particularly with many faceted mirrors within and expected additional field of view corrector makes Tkaczyk et al.'s approach less attractive for practical use. Bodkin et al. in U.S. Pat. No. 8,233,148 shows improvement over U.S. Pat. No. 8,174,694 but falls short of teaching full potential of SIS with lenslet and/or pinhole array.
The invention disclosed utilizes dispersion along an oblique direction at least 2° away from the direction of rows or columns of spherical lens array (lenslet) and/or pinhole array (PA) to allow spectral signature of pixels to spread, without overlapping on to each other and over several times row (or column) spacing of pixel array formed for one spectral channel. This would make possible increasing number of spectral channels in many SIS systems without increase in design complexity, optics and mechanical performance requirements, and parts count over comparable prior art systems. The said parameters may even be relaxed in some instances.
In an embodiment of a SIS system the lenslet is placed at or near the input to divide the field of view into a 2D array, meaning, an array of sharply focused illumination spots of each EMR input image pixel. Each lens in the lenslet array integrates image within the said lens aperture, focuses or images to a sharp point and creates nearly illumination free region surrounding it. This effectively sets the spatial resolution of the system if other components, such as FPA, input optics or collimating-reimaging optics do not restrict the same. A reflective or absorbing PA may be placed to remove unwanted EMR either scattered or misdirected due to deficient imaging/focusing capabilities of optics. If the desired nearly illumination free regions have sufficient contrast ratio to expose the FPA, then PA may not be needed. On the other hand, if the input EMR image is sufficiently strong and the imager's spatial resolution is about same to the PA row/column spacing then PA alone may work without use of lenslet. A dispersing and reimaging optics that follows directs dispersed pixels to distinct detectors in FPA. The dispersion angle is chosen to disperse the spectrum of each pixel after lenslet and/or PA not only in the dark area near the said pixel at one wavelength but in dark areas extending many times rows (or column) spacing of pixel array at single wavelength. Long path lengths of spectral signatures of dispersed pixels in the said dark region make more spectral channels possible. The spatial image on FPA of one spectral channel would coincide with lenslet and/or PA array centers following transformation according to dispersion direction and reimaging optics. The dispersed spatial image at FPA would follow the dispersion direction in the same array format with successive spectral channel shifting one pixel in the direction of dispersion. Many spectral channels would be possible before array's spectrum overlap with oblique dispersion direction.
Dispersion in oblique direction to rows or columns directions produces final image geometry at FPA with closer to 1:1 aspect ratio. Common FPAs have rectangular shape with about equal sides in general. With dispersion in oblique direction, detectors near boundaries in a rectangular shaped FPA may not receive any EMR because of the staggering of dispersed pixels at FPA. Those detector sites would not be of any use. The percentage loss of detectors for such site though undesirable, would be small and shall not be of much concern. It will be apparent to one with ordinary skill in the art that a custom FPA can be made to optimally use the characteristics of oblique dispersion.
In other embodiments of SIS systems, lenslet directly focuses each individual input pixel on to a unique detector of the FPS. The lenslet has integral properties of oblique dispersion and focusing. These embodiments have an advantage of fewer optical components. PA may not be used in one or more of these embodiments.
It is understood that dispersive property can be obtained with a wedge shaped optical element having EMR dispersion or with a grating, blazed or un-blazed. Therefore, a dispersive wedge can be substituted for grating or vice versa in any of the embodiments illustrated. The FPA generates radiation intensity value of each pixel in the array. The values are read out at a specific time in a buffer for processing. The FPA pixel positions are re-formatted according to the system design and a data cube is stored in the memory for later usage or display in real time.
Description that follows provides detailed information of each embodiment of the invention with reference to figures listed in BRIEF DESCREPTION. It is understood that while avoiding description of generally known features for brevity, many specific details and examples are given here to provide easy understanding of the concepts but the invention may be practiced without such specific details as apparent to one of ordinary skill in the art.
Lenses 2110 in the lenslet and focused array of pixes 2201 from lenslet and/or PA are depicted in the inset with column spacing Ax and row spacing Ay. The focused or imaged array of pixels 2201 from lenslet-PA are collimated by a collimating lens 2300 and dispersed at an oblique angle 0 from lenslet-PA rows or columns direction by a dispersing element 2500. The said element 2500 depicts wedge in both x and y direction. 0 is chosen equal or greater than 2° to obtain non-overlapping spectral signature spreading length of pixels longer than several times row (or column) spacing of a pixel array formed for one spectral channel. This is schematically shown in
For M×N resolvable spatial field and S spectral channels at least M×N×S number of detectors would be needed within a FPA. Since an array of M×N pixels form a spatial image at one spectral channel and when shifted one detector position in oblique direction represent next spectral channel, M×N×S active detectors are needed. The obliquely dispersed pixel array has staggered rectangular shape. Therefore, a rectangular FPA would require some additional detectors lying near the edges that may not receive useful EMR. However, the fraction of such inactive detectors would be small within the FPA.
FPAs with over 50 Megapixels have become available and may be utilized for high spatial and spectral resolution SIS systems. Those FPAs may provide high resolution systems such as an example, 200 spectral channels and 500×500 spatial pixels image. Prior art embodiments would be impractical for such high resolution systems, while systems utilizing oblique dispersion and lenslet and/or PA as disclosed would be more feasible.
In another embodiment in which lenslet's and front imager's optical quality is superior so that illumination in the regions 2202 is sufficiently low for FPA response; SIS systems may be built without PA 2200 in this case as shown in
A blazed or un-blazed grating can perform dispersive function in place of optical wedge. Etching or micromachining of grooves on the focusing side of lenslet may be easier in some instances.
A self-focusing fiber section known as “green lens” may be employed as lens in the lenslet. Green lenses provide superior focusing and imaging quality and can be fabricated in long bundle form and may be sectioned to form lenslet. A lenslet 2102 using green lenses 2114 is illustrated in
It is understood that while the invention has been described using few exemplary embodiments, those with ordinary skill in the art and with the knowledge of this disclosure may devise other methods and other embodiments to achieve the same without departing from the scope of the invention as disclosed herein. Therefore, the scope of this invention is limited by the following claims only.
This application claims priority to U.S. Provisional Application No. 61/911,108 filed on Dec. 3, 2013.
Number | Date | Country | |
---|---|---|---|
61911108 | Dec 2013 | US |