The non-uniform response of infrared focal plane array (IRFPA) detectors has a serious impact on image quality. First, the image is not clear visually, there are serious fixed mode noise, and second, the dynamic range, which reduces the resolution of infrared image. In view of the shortcomings of the existing non-uniformity correction schemes and the actual engineering requirements, this paper proposes a scheme combining registration-based correction algorithm with micro-motion system, which calculates the relative displacement between two adjacent images by projection estimation. At this time, the scene motion is judged. When the motion is inadequate, the scene motion is controlled to make the controller operate according to a certain trajectory, so that the scene conditions can meet the requirements of the correction algorithm as far as possible, solve the practical application requirements, make the image quality achieve better results, and perform the performance of the algorithm. The test results show that the scheme can effectively correct the non-uniformity and obtain better correction effect.
Deep learning models have demonstrated great success in various computer vision tasks such as image classification and object tracking. However, tracking the lumbar spine by digitalized video fluoroscopic imaging (DVFI), which can quantitatively analyze the motion mode of spine to diagnose lumbar instability, has not yet been well developed due to the lack of steady and robust tracking method. In this paper, we propose a novel visual tracking algorithm of the lumbar vertebra motion based on a Siamese convolutional neural network (CNN) model. We train a full-convolutional neural network offline to learn generic image features. The network is trained to learn a similarity function that compares the labeled target in the first frame with the candidate patches in the current frame. The similarity function returns a high score if the two images depict the same object. Once learned, the similarity function is used to track a previously unseen object without any adapting online. In the current frame, our tracker is performed by evaluating the candidate rotated patches sampled around the previous frame target position and presents a rotated bounding box to locate the predicted target precisely. Results indicate that the proposed tracking method can detect the lumbar vertebra steadily and robustly. Especially for images with low contrast and cluttered background, the presented tracker can still achieve good tracking performance. Further, the proposed algorithm operates at high speed for real time tracking.
Super-resolution is being considered as one of the important goals for optical imaging and image processing. In this paper, we present a novel imaging technique that exceeds the limit of resolving power of diffraction-limited optical system to achieve super-resolution imaging, by combining the advantages of compressive sensing and complex annular filters. This technique is realized by utilizing a classical 4F optical system with a phase-only spatial light modulator. Furthermore, the feasibility of this technique is theoretically analyzed, and physically validated by laboratory experiments. Experimental results demonstrated that this technique improves the resolving power of diffraction-limited optical system by approximately 1.58 times, and the intensity image of high-resolution object can be recovered from 25% of the total number of measurements.
With the latest improvements of microbolometer focal plane arrays (FPA), uncooled infrared (IR) cameras are becoming the most widely used devices in thermography, especially in handheld devices. However the influences derived from changing ambient condition and the non-uniform response of the sensors make it more difficult to correct the nonuniformity of uncooled infrared camera. In this paper, based on the infrared radiation characteristic in the TEC-less uncooled infrared camera, a novel model was proposed for calibration-based non-uniformity correction (NUC). In this model, we introduce the FPA temperature, together with the responses of microbolometer under different ambient temperature to calculate the correction parameters. Based on the proposed model, we can work out the correction parameters with the calibration measurements under controlled ambient condition and uniform blackbody. All correction parameters can be determined after the calibration process and then be used to correct the non-uniformity of the infrared camera in real time. This paper presents the detail of the compensation procedure and the performance of the proposed calibration-based non-uniformity correction method. And our method was evaluated on realistic IR images obtained by a 384x288 pixels uncooled long wave infrared (LWIR) camera operated under changed ambient condition. The results show that our method can exclude the influence caused by the changed ambient condition, and ensure that the infrared camera has a stable performance.
This paper proposes a novel imaging technique which combines clustering sub-dictionary learning and gradient histogram preservation to improve the quality of compressive imaging from two aspects: edge sharpness and noise suppression. Practical experiments further demonstrate better results on practical optical imaging application in terms of weighted peak signal-to-noise ratio and measure of feature similarity index.
The non-uniformity response in infrared focal plane array (IRFPA) detectors has a bad effect on images with fixed pattern noise. At present, it is common to use shutter to prevent from radiation of target and to update the parameters of non-uniformity correction in the infrared imaging system. The use of shutter causes „freezing‟ image. And inevitably, there exists the problems of the instability and reliability of system, power consumption, and concealment of infrared detection. In this paper, we present an efficient shutter-less non-uniformity correction (NUC) method for infrared focal plane arrays. The infrared imaging system can use the data gaining in thermostat to calculate the incident infrared radiation by shell real-timely. And the primary output of detector except the shell radiation can be corrected by the gain coefficient. This method has been tested in real infrared imaging system, reaching high correction level, reducing fixed pattern noise, adapting wide temperature range.
In this paper we present a novel method to record high spatial resolution far-infrared(FIR) hologram. This method takes advantage of the photo-induced phase transition characteristic of vanadium dioxide(VO2) film. The light path is off-axis digital hologram recording path, while the VO2 film is kept in constant temperature in front of the recording high resolution CMOS sensor. In the setup, the far infrared light from CO2 laser changes the partial transmittance of VO2 film to visible light, then a read-out visible laser is used to measure the transmittance of VO2 film, and subsequently the results are recorded by a high resolution CMOS sensor. So that with utilizing the photo-induced phase transition of VO2 film, we can use CMOS sensor to record far-infrared digital hologram. As the pixel pitch of CMOS sensor is much smaller than tradition FIR sensor, the recorded FIR digital hologram has been much improved. Moreover, the transition speed of VO2 film is in nanosecond scale which means that far-infrared fast-moving object recording and hologram video could be achieved. In our experiments we used different objects to compare the spatial recording resolution and the experiments prove that our method can record higher spatial spatial resolution than traditional FIR digital hologram. It has the potential to become a more effective FIR digital hologram record method. Further research will focus on the simplified light path and FIR hologram video record and process.
Recent studies have shown that compressed sensing is capable of recover sparse signals with much few measurements. Meanwhile, it provides a new idea for super resolution imaging. However, previous compressed sensing super resolution imaging methods use digital micro-mirror device or coded aperture as the measurement matrix, which makes the method inefficient. In this paper, we propose a super-resolution method via parallel compressed sensing, the proposed method using scattering medium as the measurement matrix, which we can get enough measurement values at once. Each measurement values contains global information about the object. Experiment simulation results show the effectiveness of the proposed method.
Restricted by the detector technology and optical diffraction limit, the spatial resolution of infrared imaging system is
difficult to achieve significant improvement. Super-Resolution (SR) reconstruction algorithm is an effective way to solve
this problem. Among them, the SR algorithm based on multichannel blind deconvolution (MBD) estimates the
convolution kernel only by low resolution observation images, according to the appropriate regularization constraints
introduced by a priori assumption, to realize the high resolution image restoration. The algorithm has been shown
effective when each channel is prime. In this paper, we use the significant edges to estimate the convolution kernel and
introduce an adaptive convolution kernel size selection mechanism, according to the uncertainty of the convolution
kernel size in MBD processing. To reduce the interference of noise, we amend the convolution kernel in an iterative
process, and finally restore a clear image. Experimental results show that the algorithm can meet the convergence
requirement of the convolution kernel estimation.
Nonuniformity in FPAs (focal plane arrays) is a common, although undesirable, characteristic arising from small differences in the responsivity of individual detectors. Now there are many scene-based nonuniformity correction algorithms. In these correction algorithms, interframe registration based nonuniformity correction algorithm is a principal type. The core of interframe registration based nonuniformity correction algorithm is accurate image registration. However, because of the existence of the nonuniformity, the values of the adjacent columns or rows of the image with noise are so close that it is difficult to distinguish them. We find it is difficult to achieve the corresponding rows or columns registration between two frames accurately. It will lead to the wrong registration that will affect the nonuniformity correction effect. In this method, we propose a nonuniformity correction algorithm based on subspace projection registration. By projecting the rows and columns of the image on to their corresponding subspace respectively, we can realize the dimension reduction of the row space and the column space. As a result, the value differences between the adjacent rows or the adjacent columns of the image can be amplified. Then the algorithm can identify the corresponding rows and columns between two frames, enabling accurate image registration. Experiments demonstrate that our algorithm achieve accurate image registration and outstanding correction effect.
Colorized linear cameras deliver superb color fidelity at the fastest line rates in the industrial inspection. It’s RGB
trilinear sensor eliminates image artifacts by placing a separate row of pixels for each color on a single sensor. It’s
advanced design minimizes distance between rows to minimize image artifacts due to synchronization. In this paper, the high-speed colorized linear CCD data acquisition system was designed take advantages of the linear CCD sensor
μpd3728. The hardware and software design of the system based on FPGA is introduced and the design of the functional modules is performed. The all system is composed of CCD driver module, data buffering module, data processing module and computer interface module. The image data was transferred to computer by Camera link interface. The system which automatically adjusts the exposure time of linear CCD, is realized with a new method. The integral time of CCD can be controlled by the program. The method can automatically adjust the integration time for different illumination intensity under controlling of FPGA, and respond quickly to brightness changes. The data acquisition system is also offering programmable gains and offsets for each color. The quality of image can be improved after calibration in FPGA. The design has high expansibility and application value. It can be used in many application situations.
Combining with the current development trend in video surveillance-digitization and high-definition, a multimode-compatible image acquisition system for HD area array CCD is designed. The hardware and software designs of the color video capture system of HD area array CCD KAI-02150 presented by Truesense Imaging company are analyzed, and the structure parameters of the HD area array CCD and the color video gathering principle of the acquisition system are introduced. Then, the CCD control sequence and the timing logic of the whole capture system are realized. The noises of the video signal (KTC noise and 1/f noise) are filtered by using the Correlated Double Sampling (CDS) technique to enhance the signal-to-noise ratio of the system. The compatible designs in both software and hardware for the two other image sensors of the same series: KAI-04050 and KAI-08050 are put forward; the effective pixels of these two HD image sensors are respectively as many as four million and eight million. A Field Programmable Gate Array (FPGA) is adopted as the key controller of the system to perform the modularization design from top to bottom, which realizes the hardware design by software and improves development efficiency. At last, the required time sequence driving is simulated accurately by the use of development platform of Quartus Ⅱ12.1 combining with VHDL. The result of the simulation indicates that the driving circuit is characterized by simple framework, low power consumption, and strong anti-interference ability, which meet the demand of miniaturization and high-definition for the current tendency.
This paper presents a high‐definition video display solution based on the FPGA and THS8200. THS8200 is a video decoder chip launched by TI company, this chip has three 10-bit DAC channels which can capture video data in both 4:2:2 and 4:4:4 formats, and its data synchronization can be either through the dedicated synchronization signals HSYNC and VSYNC, or extracted from the embedded video stream synchronization information SAV / EAV code. In this paper, we will utilize the address and control signals generated by FPGA to access to the data‐storage array, and then the FPGA generates the corresponding digital video signals YCbCr. These signals combined with the synchronization signals HSYNC and VSYNC that are also generated by the FPGA act as the input signals of THS8200. In order to meet the bandwidth requirements of the high‐definition TV, we adopt video input in the 4:2:2 format over 2×10-bit interface. THS8200 is needed to be controlled by FPGA with I2C bus to set the internal registers, and as a result, it can generate the synchronous signal that is satisfied with the standard SMPTE and transfer the digital video signals YCbCr into analog video signals YPbPr. Hence, the composite analog output signals YPbPr are consist of image data signal and synchronous signal which are superimposed together inside the chip THS8200. The experimental research indicates that the method presented in this paper is a viable solution for high‐definition video display, which conforms to the input requirements of the new high‐definition display devices.
In recent years based on security, quality supervision, inspection and medical for the urgent need of infrared temperature
measurement and infrared display technology, coupled with embedded system to achieve rapid development, which is
widely used in the electronic products and the field of intelligent instruments and industrial control, this paper has
designed a kind of more comprehensive, more efficient and more intuitive infrared thermometer. Unlike previous
handheld infrared thermometer, we regard an embedded Linux system as the system, with its open source code, support
most mainstream hardware platforms, unified peripheral interface and can be customized, to build an embedded infrared
system that has provided strong system support; the pseudocolor techniques and Qt interface display technology make
the image more colorful and the picture function more diverse; With ARM microprocessor as the display and
temperature measuring platform, it costs reduction and reduce volume and power consumption; the FrameBuffer
interface technology and multithreading technology realize the smooth real-time display. And ultimately the display size
of real-time infrared image is 640 * 480 at a speed of 25 frames / sec. What is more, display is equipped with the menu
option so that thermometer can be required to complete the operation through the button. The temperature display system
aims at small volume, easy to use and flexible. I believe this thermometer will have a good application prospect.
Dynamic range reduction and detail enhancement are two important issues for effectively displaying high-dynamic-range images acquired by thermal camera systems. They must be performed in such a way that the high dynamic range image signal output from sensors is compressed in a pleasing manner for display on lower dynamic range monitors without reducing the perceptibility of small details. In this paper, a new method of display and detail enhancement for high dynamic range infrared images is presented. This method effectively maps the raw acquired infrared image to 8-bit domain based on the same architecture of bilateral filter and dynamic range partitioning approach. It includes three main steps: First, a bilateral filter is applied to separate the input image into the base component and detail component. Second, refine the base and detail layer using an adaptive Gaussian filter to avoid unwanted artifacts. Then the base layer is projected to the display range and the detail layer is enhanced using an adaptive gain control approach. Finally, the two parts are recombined and quantized to 8-bit domain. The strength of the proposed method lies in its ability to avoid unwanted artifacts and adaptability in different scenarios. Its great performance is validated by the experimental results tested with two real infrared imagers.
A Mean-shift Particle filtering tracking algorithm based on the multi-feature fusion has been raised in this paper. This
algorithm mainly focus on the features of the high frequency histogram, fractal and the energy of the infrared small target,
which directly against the defects exist in detecting the infrared small targets, such as the size of the target, the low
tracking accuracy caused by the low SNR and so on. Since the particle filtering algorithm gives the advantage of
multi-feature fusion, the algorithm raised in this paper combines the three features listed above and does the calculation
using the particle weight to greatly improved the tracking accuracy. The clustering effect of the Mean-shift algorithm has
also been applied to make the distribution of the particles more equals to the real target, which reduced the number of the
particle and enhanced the real-time ability of the algorithm. The experimental results show that, this algorithm has better
tracking accuracy, which gives more effectiveness in tracking the infrared small target compared to the traditional
particle filtering algorithm.
In scene-based nonuniformity correction (NUC) methods for infrared focal plane array cameras, the statistical approaches have been well studied because of their lower computational complexity. However, when the assumptions imposed by statistical algorithms are violated, their performance is poor. Moreover, many of these techniques, like the global constant statistics method, usually need tens of thousands of image frames to obtain a good NUC result. In this paper, we introduce a new statistical NUC method called the multiscale constant statistics (MSCS). The MSCS statically considers that the spatial scale of the temporal constant distribution expands over time. Under the assumption that the nonuniformity is distributed in a higher spatial frequency domain, the spatial range for gain and offset estimates gradually expands to guarantee fast compensation for nonuniformity. Furthermore, an exponential window and a tolerance interval for the acquired data are introduced to capture the drift in nonuniformity and eliminate the ghosting artifacts. The strength of the proposed method lies in its simplicity, low computational complexity, and its good trade-off between convergence rate and correction precision. The NUC ability of the proposed method is demonstrated by using infrared video sequences with both synthetic and real nonuniformity.
Research of image processing algorithm is one of the key issues for infrared imager development. Nowadays, researchers
of image processing algorithm and designers of infrared imager have presented many image processing algorithms. It is
necessary to evaluate the practicability, real-time performance and adaptability of all these algorithms in advance of
application. Based on virtual instrumental technology, a general demo and evaluation system for infrared image
processing algorithms is developed. The system configuration is described in detail. The extendable property of this
system made it suitable for various algorithms demo and evaluation. The vision impression of image processing
algorithms demo is processed through labview programming. Designers can evaluate the performance of image
processing algorithms in time. This system benefits designers to optimize their algorithms directly. Examples are applied
in this system to prove its functions. Trial results show it is a useful tool for infrared imager developer and image
processing algorithm designer.