Studies On Instant Sea Cucumber Freshness Detection Based On The Fusion Of Mechanical Properties And Visual Textures

Because of the unique physical structure,instant sea cucumber is extremely easy to spoil during the storage process in the natural environment,corruption of instant sea cucumber not only reduce its quality,but also cause food safety problems.So it has a great significance to detection the freshness of instant sea cucumber.Currently,there are generally two ways to detect foods freshness,such as sensory detection,physical and chemical detection.The result of sensory detection is easy to influence by subjective factors.Physical and chemical detection also has some disadvantages,such as tedious detection process,more time-consuming,and need dedicated analytical instruments.In the present study,to realize the automation division of freshness of sea cucumber,assessment model was build through Fuzzy Neural Network based on the tactile and vision features characteristic.The main research contents were as follows:1.A mechanical properties detection device was designed.The 3D modeling of this device was constructed by Solidworks.To achieve the action of contact sea cucumber,motor unit was build through selecting the appropriate motor and control card.To measure the pressure that sea cucumber forces on the proper and remove the interference data during the contacted progress,data processing unit was build.In the end,the control software of this device was programmed.2.By scanning electron microscopy experiments,the regularity of internal organizational structure with the instant sea cucumber freshness change was observed,and the relevance between freshness and mechanical properties in mechanism aspects was verified.In addition,through detecting the mechanical properties of instant sea cucumber,a mechanical characteristic curve was established,by comparing the curve between different freshness of instant sea cucumber,the maximum pressure FN,the counting ration(the P value)and the elastic potential energy W were adopt as the mechanical properties which can reflect the freshness of instant sea cucumber.3.Through the analysis of different freshness of instant sea cucumber image,get the change of texture characteristics with fresh degree.Then a machine vision system was build to collect the image of sea cucumbers,and the image was processed by the method of gray scale and median filtering.Therefore,a grey-gradient co-occurrence matrix was constructed to extract texture feature value.Subsequently,by analyzing the correlation between each characteristic values and using the principal component analysis method to reduce the dimension of texture feature,two principal components which can reflect the texture information of sea cucumber was extracted.4.On the basis of analyzing the structure of fuzzy neural network,two evaluation models of sea cucumber based on the mechanical properties and the visual features were established,the result of this two models were 76.2%and 73.8%.Through analyzing the test result of this two network,an improved model which fusion mechanical properties and visual features was established,the test result show that the accuracy of this model is 94.04%.