Automatic detection and recognition of Indian currency note has gained a lot of research attention in recent years particularly due to its vast potential applications. In this paper we introduce a new recognition method for Indian currency using computer vision. It is shown that Indian currencies can be classified based on a set of unique non discriminating features such as color, dimension and most importantly the Identification Mark (unique for each denomination) mentioned in RBI guidelines. Firstly the dominant color and the aspect ratio of the note are extracted. After this the segmentation of the portion of the note containing the unique I.D. Mark is done. From these segmented image, feature extraction is done using Fourier Descriptors. As each note has a unique shape as the I.D. Mark, the classification of these shapes is done with the help of Artificial Neural Network. After feature extraction, the denominations are recognized based on the developed algorithm. The success rate of the proposed system is 97% requiring a processing time of 2.52 seconds.

No abstract

No abstract

The manuscript proposes a novel architecture of a delay cell that is implemented in 4-stage VCO which has the ability to operate in two distributed frequency bands. The operating frequency is chosen based on the principle of carrier mobility and the transistor resistance. The VCO uses dual delay input techniques to improve the frequency of operation. The design is implemented in Cadence 90nm GPDK CMOS technology and simulated results show that it is capable of operating in dual frequency bands of 55 MHz to 606 MHz and 857 MHz to 1049 MHz. At normal temperature (270) power consumption of the circuit is found to be 151?W at 606 MHz and 157?W at 1049 MHz respectively and consumes an area of 171.42?m2. The design shows good tradeoff between the parameters-operating frequency, phase noise and power consumption.

Abstract-Digital watermarking plays a very important role in copyright protection. It is one of the techniques which are used for safeguarding the origins of the image, audio and video by protecting it against Piracy. This paper proposes a low variance based spread spectrum watermarking for image and video in which the watermark is obtained twice in the receiver. The watermark to be added is a binary image of comparatively smaller size than the Cover Image. Cover Image is divided into number of 8x8 blocks and transform into frequency domain using Discrete Cosine Transform. A gold sequence is added as well as subtracted in each block for each watermark bit. In most cases, researchers has generally used algorithms for extracting single watermark and also it is seen that finding the location of the distorted bit of the watermark due to some attacks is one of the most challenging task. However, in this paper the same watermark is embedded as well as extracted twice with gold code without much distortion of the image and comparing these two watermarks will help in finding the distorted bit. Another feature is that as this algorithm is based on embedding of watermark in low variance region, therefore proper extraction of the watermark is obtained at a lesser modulating factor. The proposed algorithm is very much useful in applications like realtime broad casting, image and video authentication and secure camera system. The experimental results show that the watermarking technique is robust against various attacks.