Search

Stable imaging in adverse environments (e.g., total darkness) makes thermal infrared (TIR) cameras a prevalent option for night scene perception. However, the low contrast and lack of chromaticity of TIR images are detrimental to human interpretation and subsequent deployment of RGB-based vision algorithms. Therefore, it makes sense to colorize the nighttime TIR images by translating them into the corresponding daytime color images (NTIR2DC). Despite the impressive progress made in the NTIR2DC task, how to improve the translation performance of small object classes is under-explored. To address this problem, we propose a generative adversarial network incorporating feedback-based object appearance learning (FoalGAN). Specifically, an occlusion-aware mixup module and corresponding appearance consistency loss are proposed to reduce the context dependence of object translation. As a representative example of small objects in nighttime street scenes, we illustrate how to enhance the realism of traffic light by designing a traffic light appearance loss. To further improve the appearance learning of small objects, we devise a dual feedback learning strategy to selectively adjust the learning frequency of different samples. In addition, we provide pixel-level annotation for a subset of the Brno dataset, which can facilitate the research of NTIR image understanding under multiple weather conditions. Extensive experiments illustrate that the proposed FoalGAN is not only effective for appearance learning of small objects, but also outperforms other image translation methods in terms of semantic preservation and edge consistency for the NTIR2DC task. Compared with the state-of-the-art NTIR2DC approach, FoalGAN achieves at least 5.4% improvement in semantic consistency and at least 2% lead in edge consistency.

Detecting of the auto parts solder joints quality is always a technological difficulty in the production line, the key point is how to exact the parts feature which is tested. New quality detection method for the auto parts soldered joints is presented based on biology vision feature. First the Gabor filter banks which have the emulation of biological vision is used to filter the detected image with eight different directions. Then the biggest powers of all pixel points are chosen to be the solder joints feature and classified by support vector machine (SVM). The experimental results show that the algorithm has higher accuracy as an effective quality detection method. It can be a useful reference to the auto parts solder joints quality testing of engineering application.

Aim at high-to-float high-ash fine coal with large content of granules, in order to reduce pollution of carried fine slime to quality of concentrate coal, soluble glass, sodium hexametaphosphate and cornstarch these three kinds of regulators were choosen to carry out flotation reducing ash experiment. Result showed that, cause of poor selectivity of fine coal is mainly small quality, big surface area and the poor dispersion of fine coal. When sodium hexametaphosphate dosage is 1500g/t, the kerosene dosage is 370g/t, the foam beater GF dosage is 150g/t, production rate of concentrate coal is 57.69%，and ash content of concentrate coal is 12.04%. Under the same condition, separation index of sodium hexametaphosphate as regulator is superior to cornstarch and soluble glass.

As for certain high ash and difficult floatable fine coal, the coal quality characteristic is the frangible and easyly-clayed main gangue mineral kaolinite, and the high content of fine-grade and intermediate density grade materials. In order to reduce tiny slimes pollution and to increase quality and recovery rate of concentrate, we used cyclonic-static micro-bubble flotation column and XFD-3 flotation machine to experiment on separation process. It probed into re-concentration after grinding rough refined coal, classified flotation process and selective agglomeration-flotation process these three technical processes. The comparison of flotation machine test result showed that, using the flotation column as separation equipment, process of re-concentration after grinding rough refined coal could obtain the best refined coal accumulation index. When medicaments dosage is kerosene 800 g/t, GF 300 g/t, and sodium hexametaphosphate 1500g/t, refined coal production rate of flotation column is 59.47% and ash content is 10.63%.

As certain refractory molybdenum with degree high oxidation and high argillization, on the premise of molybdenum concntrate recovery rate, in order to increase tenor of molybdenum concntrate, we use cyclonic static micro-bubble flotation column as concentration equipment. We used two concentrate courses to replace flotation machine eight concentrate courses. It showed that, in the condition of same recovery rate, tenor of molybdenum concntrate was 26.16%, rising by 8%. Using processes of flotation machine eight concentrate courses - concentrate regrinding - flotation column two concentrate courses, tenor of molybdenum concntrate was 39.33% and the recovery rate was of 8%. The results showed that, by increasing the degree of monomer dissociation of molybdenum ores, cyclonic static micro-bubble flotation column has higher capability in recovering and selecting ores of -400 mm size fraction. The high argillization resulting from the regrinding and unsuppressed gangue minerals are the main factors affecting the tenor of ore concentrate.

In this paper, carbon materials with solid and hollow sphere morphologies, have been prepared by reduction of hexachlorobenzene with metallic sodium at temperatures of 400-500°C for 6-10h. Characterized by XRD, TEM, SAED and Raman spectrum, the composition and morphology of the products are confirmed. The possible formation mechanism has also been proposed.