Based on the traditional particle swarm search algorithm, a new search algorithm is proposed, which is called Adaptive Decision Inertia weight Particle Swarm Optimization (ADI-PSO) correlation searching algorithm. ADI-PSO combines adaptive inertia factor and policy termination, which can continually update inertia factor to accelerate the convergence speed during the iterations. The algorithm can intelligently determine whether to end the iteration, so it can get accurate search results more quickly. Two groups of contrast experiments show that the precision of ADI-PSO algorithm and N-R algorithm are much the same, but the search speed of ADI-PSO is more quickly. In the case of rough speckle image, the optimal fitness value is -0.038 6, 0.888 1, 0.917 6 respectively using three different methods. Compared with PSO and LDI-PSO, the convergence speed of ADI-PSO is more quickly, and the stability is better. Besides, ADI-PSO can overcome premature convergence effectively.

Stress inspection of large diameter circular saw can be accomplished by the deformation measurement, and the result will fluctuate because of the vibration of the saw in the detection process. In order to improve the accuracy of measurement, a detecting algorithm was presented based on curve fitting. The fixed size sliding window method was used to reconstruct the quantitative data, and a new data structure was formed. The detection noise can be decreased by curve fitting of the data structure. The deformation inspection system was developed and a 1 700 mm diameter sample was measured. Results indicate that the proposed algorithm reduces the deformation errors brought by detection system itself and vibration of circular saw blade. The measurement precision is up to 0.02 mm.

In order to realize the precise measurement of flatness for large diameter narrow zone annular plane, a novel method based on optics and image processing using inner focus collimator and precision rotary table is proposed. Firstly, the image of point light source is projected above the annular plane using inner focus collimator. And the image point is driven by precision rotary table to draw a circle. The circle plane is the reference. The point light source is imaging on the CCD. Then the distance between annular plane and reference plane is measured. Next, the testing data is analyzed by means of Fourier harmonic analysis method. The influence of constant item and the first item is removed. Then the flatness of the annular plane can be calculated by the data processing. The basic principle of the testing method is analyzed. An annular workpiece with the out-diameter of 5 m and the inner-diameter of 4.4 m and the height of 0.5 m is tested with this method. The flatness is 0.285 mm, and the repeatability is 0.009 3 mm. Finally, the measurement uncertainty of this method is analyzed, and the composed standard uncertainty is 0.007 4 mm.

Three-dimensional (3D) shape measuring techniques are based on the principle of triangulation and technique of fringe projection. In the process of measuring, triangular projection often produces local shadow and occlusion. In addition, some dark marks are needed to be attached to the tested surface for convenient later data merging. These may lead to local fracture of fringe pattern and finally lose part point-cloud data on the surface of tested objects. In order to ensure the integrity and accuracy of 3D shape measurement results, Radial Basis Function (RBF) is applied to repair the missing 3D point-cloud data. The results of simulation and actual experiment show that the surface interpolated by the algorithm in this article is smooth, which is closed to the original data. And this method is expected to repair missing data in the process of 3D measurement.

In order to upgrade the cut-by-length technology of billet to cut-by-weight, a real-time accumulation vision weight method of billet based on linear structured laser was proposed, and its principle, measurement model and weighing process were studied. According to the vision measurement principle of linear structure laser, the current cross section area of billet was measured and calculated by two vision sensors. Meanwhile, the distance between the current cross section and the previous one was acquired by another vision sensor, and the billet weight between two adjacent cross sections was calculated and accumulated. Thus, the billet weight passing the measuring position was monitored real-timely, and the real-time weight-cut control of billet was realized. The experimental system was set up and the weight experiment was carried out with rectangular solid steel. Experimental results show that the relative error of weight is 1‰, which is far less than 8‰ used in present production enterprises. This research provides a theoretical reference for the realization of billet cut-by-weight technology.

In order to improve the measurement precision of the triangulation measurement system based on Synchronized Scanner, the system’s parameters are analyzed. First, the system’s principle and parameters are described briefly. Then, according to the system’s parameters, the influence on the precision from the position errors and angle errors of M1, M2, M3, M4, CCD and the lens’s position error and M3’s thick error are analyzed. Based on the influence on measurement precision, the parameters are ranked in the aspect of importance. Finally, the system is constructed and the 3D results are acquired.

Traditional thermal emission polarization model ignores the microscopic distribution, and taking this into consideration to expand modeling and simulation research of infrared thermal emission polarization characteristics. Model of the degree of polarization of the infrared thermal emission is built based on the microfacet polarized Bidirectional Reflectance Distribution Function (pBRDF) and the microstructure distribution. Utilizing the difference between the parallel and perpendicular components of the emissivity of the infrared thermal emission divided by the sum of them represent the degree of polarization. And the relationship between emissivity and hemispherical reflectance is obtained according to the Kirchhoff's law and the microfacet BRDF. Polarizing the microfacet BRDF to calculate the directional hemispherical reflectance and then directional emissivity matrix, and last to get the degree of polarization. The simulation results show that the curves of the relationship between the polarization and the emission angle of the infrared thermal emission are consistent with the actual measurement results, and compared with the simulation results of traditional thermal emission polarization model, is closer to the measured value, especially in the case of the surface roughness is larger, can more accurately reflect the infrared thermal emission polarization characteristics.

Multi-view video plus depth (MVD) is one of the main representations of 3D scene. Under the framework 3D-HEVC, the depth video intra coding has high computational complexity. A region segmentation-based fast CU size decision and mode decision algorithm for 3D-HEVC depth video intra coding is proposed. Firstly, the depth map is divided into four regions based on the edge extraction of depth video and texture detection of the corresponding color video texture. Then, the effects of depth video distortion on rendered view quality and mode distribution of intra prediction coding are statistically analyzed in each region. Finally, different CU size, rough prediction mode and DMMs decision are designed for encoding different regions. Experimental results show that the proposed algorithm save the encoding time of MVD by 55.1% on average, and the encoding time of depth video by 61.57% on average with negligible rendered virtual view image degradation.

Aiming at the problem of only estimating disparity from left and right stereo images, a novel method of stereo matching is proposed, which decomposes object images into a set of sub-bands by discrete wavelet transform, then modifies each sub-band image, and improves the matching quality of each sub-band image by implementing discrete inverse wavelet transform. On this basis, we estimate disparity of each modified image, and merge different sub-band disparity maps into a final disparity map using HVS model. Experimental results demonstrate that, compared with the existing algorithm, the proposed algorithm achieve a high-precision.

The contact between distortion of visual perception characteristics and quality of stereo image is explored, and a new stereo image quality assessment model is proposed in combination with visual perception characteristics and random forest learning method. Firstly, according to human visual system, a series of visual perception characteristics are extracted as features to measure the degree of distortion stereo image, called perception characteristics sets. Secondly, random forest learning method is utilized to simulate HVS mechanism and fuse these features, and then a model is built for evaluating stereo image quality. By applying the proposed model to stereo image test database, experimental results demonstrate that proposed model is consisted with human perception well.

As active contours without edges (C-V) model is difficult to segment thyroid ultrasound image with intensity inhomogeneity. Therefore, the improved C-V ultrasound image segmentation model of combining local information is proposed. First, the local information is not affected by the gray distribution. So, through this characteristic, we constructed a new speed function by using local image fitting information. According to the change of local gray level, the speed function can flexibly control curve evolution rate. Then, the speed function was incorporated into the C-V model, and had the ability of global segmentation. The experiment results demonstrate that the proposed model can achieve accurate segmentation for thyroid tumor ultrasound image with intensity inhomogeneity. And the segmentation efficiency is also improved.

To solve the problems such as pixel value saturation and low brightness in the result of Dark channel prior haze removal algorithm, we proposed “Airlight Map” instead of the airlight value in the original algorithm. That is to say, we used different airlight values for each pixel of the image during dehaze process. Thus the problem caused by using only one airlight value in the whole image could be solved. We assumed that the directions of brightness change in one image can be classified into four basic directions, and the airlight value was piecewise monotone along the direction of brightness change. Therefore, we generated the Airlight Map using piecewise linear interpolation. The proposed method is simple in principle, and the dehaze time dose not increase notably. At the same time, better haze removal results are obtained using our method.

Similar pixels or image patches exist in different position inside an image or between different images, so that the fusion of similar patches can reconstruct the high resolution image. In order to mine the potential similar information further for the better reconstruction quality, the paper utilizes of the geometry transformation self-similarity of images, by extracting Zernike moments features within the image neighborhood to match and measure the nonlocal similar patches among the images. Then a super-resolution reconstruction algorithm for image sequences is proposed, through the collaborative weighting fusion of similar patches with geometric transformation. The experimental results prove the validity of the proposed method and its superiority in reconstruction effect.

The gas temperature measurement was achieved by using wavelength-modulation spectroscopy technique based on the Tunable Diode Laser Absorption Spectroscopy (TDLAS). Considering that H2O exists in the atmosphere and industrial production, we choose H2O as the carrier gas of temperature measurements. Using the absorption lines near 1 397 nm, according to analysis the relationship between the second harmonic signal peak - peak ratio and temperature, temperature measurements were obtained. Experimental results show that the max error does not exceed 0.113% within the temperature range of 400 ~1 000℃. Combined with the special structure, the device can be used for temperature monitoring of industrial production process. The range of temperature monitoring instrument, the monitoring precision and accuracy is far better than the currently used. The temperature monitoring can provide accurate data for the production equipment of boiler combustion, so as to realize the production efficiency and accurate control, and achieve the ultimate goal of energy saving and emission reduction.

Pointing at the problem of the all-reflection Fourier transform spectrometer that its ability of weak light detection is wasted, we has carried a further structure on the original Fresnel double-sided mirror structure. The former coupled Fresnel double-sided mirror is separated for a distance. Meanwhile, move the two separated plane mirror up for a proper distance around the optical axis. The improved structure’s interference field is just distributed above the optical axis which was used to around the axis of symmetry and its upper and lower was symmetry. Simulations results show that the further structure’s incident light energy’s utilization rate and the interference fringe Fourier transform’s resolution are doubled. The improvement has important significance to enhance the all-reflection Fourier transform imaging spectrometer‘s weak light detection capability and expand its spectrum detection range.