Search

Selective carbon dioxide photoreduction to produce formic acid was achieved under visible light irradiation using water molecules as electron donors, similar to natural plants, based on the construction of a Z-scheme light harvesting system modified with a Cu-Zn alloy nanoparticle co-catalyst. The faradaic efficiency of our Z-scheme system for HCOOH generation was over 50% under visible light irradiation.

Photochemical removal of NO(2) in N(2) or air (5-20% O(2)) mixtures was studied by using 172-nm Xe(2) excimer lamps to develop a new simple photochemical aftertreatment technique of NO(2) in air at atmospheric pressure without using any catalysts. When a high power lamp (300 mW/cm(2)) was used, the conversion of NO(2) (200-1000 ppm) to N(2) and O(2) in N(2) was >93% after 1 min irradiation, whereas that to N(2)O(5), HNO(3), N(2), and O(2) in air (10% O(2)) was 100% after 5s irradiation in a batch system. In a flow system, about 92% of NO(2) (200 ppm) in N(2) was converted to N(2) and O(2), whereas NO(2) (200-400 ppm) in air (20% O(2)) could be completely converted to N(2)O(5), HNO(3), N(2), and O(2) at a flow rate of 1l/min. It was found that NO could also be decomposed to N(2) and O(2) under 172-nm irradiation, though the removal rate is slower than that of NO(2) by a factor of 3.8. A simple model analysis assuming a consecutive reaction NO(2)-->NO-->N+O indicated that 86% of NO(2) is decomposed directly into N+O(2) and the rest is dissociated into NO+O under 172-nm irradiation. These results led us to conclude that the present technique is a new promising catalyst-free photochemical aftertreatment method of NO(2) in N(2) and air in a flow system.

In integrated segmentation and recognition of character strings, the underlying classifier is trained to be resistant to noncharacters. We evaluate the performance of state-of-the-art pattern classifiers of this kind. First, we build a baseline numeral string recognition system with simple but effective presegmentation. The classification scores of the candidate patterns generated by presegmentation are combined to evaluate the segmentation paths and the optimal path is found using the beam search strategy. Three neural classifiers, two discriminative density models, and two support vector classifiers are evaluated. Each classifier has some variations depending on the training strategy: maximum likelihood, discriminative learning both with and without noncharacter samples. The string recognition performances are evaluated on the numeral string images of the NIST Special Database 19 and the zipcode images of the CEDAR CDROM-1. The results show that noncharacter training is crucial for neural classifiers and support vector classifiers, whereas, for the discriminative density models, the regularization of parameters is important. The string recognition results compare favorably to the best ones reported in the literature though we totally ignored the geometric context. The best results were obtained using a support vector classifier, but the neural classifiers and discriminative density models show better trade-off between accuracy and computational overhead.

In character string recognition integrating segmentation and classification, high classification accuracy and resistance to noncharacters are desired to the underlying classifier. In a previous evaluation study, the modified quadratic discriminant function (MQDF) proposed by Kimura et al. was shown to be superior in noncharacter resistance but inferior in classification accuracy to neural networks. This paper proposes a discriminative learning algorithm to optimize the parameters of MQDF with aim to improve the classification accuracy while preserving the superior noncharacter resistance. We refer to the resulting classifier as discriminative learning QDF (DLQDF). The parameters of DLQDF adhere to the structure of MQDF under the Gaussian density assumption and are optimized under the minimum classification error (MCE) criterion. The promise of DLQDF is justified in handwritten digit recognition and numeral string recognition, where the performance of DLQDF is comparable to or superior to that of neural classifiers. The results are also competitive to the best ones reported in the literature.