Stalked protozoa identification by image analysis and multivariable statistical techniques

摘要

Protozoa are considered good indicators of thetreatment quality in activated sludge systems as they aresensitive to physical, chemical and operational processes.Therefore, it is possible to correlate the predominance ofcertain species or groups and several operational parameters ofthe plant. This work presents a semiautomatic image analysisprocedure for the recognition of the stalked protozoa speciesmost frequently found in wastewater treatment plants by determiningthe geometrical, morphological and signature dataand subsequent processing by discriminant analysis and neuralnetwork techniques. Geometrical descriptors were found to beresponsible for the best identification ability and the identificationof the crucial Opercularia and Vorticella microstomamicroorganisms provided some degree of confidence toestablish their presence in wastewater treatment plants.