On-line learning of mutually orthogonal subspaces for face recognition by image sets.

We address the problem of face recognition by matching image sets. Each set of face images is represented by a subspace (or linear manifold) and recognition is carried out by subspace-to-subspace matching. In this paper, 1) a new discriminative method that maximises orthogonality between subspaces is proposed. The method improves the discrimination power of the subspace angle based face recognition method by maximizing the angles between different classes. 2) We propose a method for on-line updating the discriminative subspaces as a mechanism for continuously improving recognition accuracy. 3) A further enhancement called locally orthogonal subspace method is presented to maximise the orthogonality between competing classes. Experiments using 700 face image sets have shown that the proposed method outperforms relevant prior art and effectively boosts its accuracy by online learning. It is shown that the method for online learning delivers the same solution as the batch computation at far lower computational cost and the locally orthogonal method exhibits improved accuracy. We also demonstrate the merit of the proposed face recognition method on portal scenarios of multiple biometric grand challenge.

An objective method to monitor and quantify texture of salted herring during ripening in barrels at 4°C

Changes in the texture (elastic nature) of the flesh of barrel salted herring during the ripening process at 4°C have been monitored. The method employs the analysis of stress-relaxation curves after compression to half of the sample thickness on an lnstron Model 1112. The parameter 'T/P' for each sample represents the reciprocal of the gradient of a line connecting P and T0.368p. This parameter characteristic of each sample's texture was calculated as the ratio of 'T/P' where, T is the relaxation time and is defined as the time required for a stress at constant strain to decrease to 1/e of its original value, where 'e' is the base of natural logarithms (2.7183). Since 1/e=0.368, the relaxation time is the time required for the force to decay to 36.8% of its original value. P is the peak height of the curve (i.e. the force value at the maximum height). This method was adopted from the bakery industry for testing the degree of gluten development in bread dough. The 'T/P' values obtained over the course of ripening for differently treated salted-herring in barrels ranged between 1 and 12. The trends in 'T/P' value, during ripening period for the different samples, appeared to be parallel changes in texture perceived by sensory observation (subjective measurement), although the heterogeneous nature of the samples gave standard deviations, about the replicate sample mean, around 5%. The method appears promising as an objective measure for monitoring this aspect of the textural quality of barrel salted-herring through ripening if reproducibility of test results can be improved by more careful standardization of sample preparation and test protocol.

Dried fish cakes with improved texture

To overcome the problem of underutilization of marine by-catches, Basu et al. (1985) developed fish cube from minced fish meat. Although the product was acceptable, it had little rubbery texture. An attempt was made to improve the texture of the cake by several methods. It was found that 5% tapioca starch along with 3% texturised soybean protein improved the texture and juiciness of the rehydrated product. Preheating of the minced meat at 70°C for 30 minutes also improved the texture appreciably. It was also found that mixing of the ingredients at low speed (less than 100 rpm) in a dough mixer gave the best texture, higher speed and sharp blades leading to rubbery texture. The dehydrated product (moisture 19-20%) thus prepared had a shelf life of six months at ambient temperature.