Study And Realization On Intellectual Detection Of Pork Fresh Degree Based On Biological Features

With the development of people's health consciousness, people's demand of diet sanitation standard is becoming higher and higher. More and more people concern about how to recognize the food freshness fastly, accurately and conveniently. Pork is one of the most popular non-staple foods in our daily life. The research of pork freshness detection has some kind of practical value. Nowadays, the detection methods of pork freshness are mainly tranditional, which include the following ones: the sensory detection, physical detection and microorganism detection etc., There are disadvantages of all the detection methods above, such as easy to be influenced by observers, complex process, long detection time and high cost.The development and application of gas detection technology, charge coupled device (CCD) technology, digital image processing technology, artificial neural network technology and digital signal processing (DSP), provide substantial theory and technology support to the study on the pork freshness. In this paper, the experimental colletion of the main feature parameters which can demonstrate the pork freshness is realized through the application of gas detection technology, CCD technology as well as digital image processing technology. The identifying of pork freshness degree was realized by making using of the artificial neural network. The hardware design of the intelligent pork freshness detection system is initially achieved on the base of DSP technology.After analyzing the mechanism of pork putridity, it was found there are complex physical changes and chemical reactions in the process of the corruption. It is hard to get accurate pork freshness degree if we simply rely on a single kind of method or data of certain aspect.It has been discovered that the gases of NH3 and H2S emit from pork during the putridity process, the change of pork color as well as the complete adipose cell number per unit area can reflect the degree of pork putrity in some extent.The experimental data and images of NH3 and H2S, color value, complete adipose cell number per unit area per different time intervals is collected in the same closed atmosphere at the same time. The experimental data of NH3 and H2S was collected through a simple gas collection system based on sensors. The experimental data of color value (H, S, I) and complete adipose cell number per unit area was realized first through image by the CCD vidicon and the image collection card. Then the images were processed with the arithmetic such as image strengthening, Hough transformation, automatic seg mentation, transform of color model, etc., After that, we can finally get the color value (H, S, I) and complete adipose cell number per unit area. TVB-N is a national standard of detecting the meat fresh degree in our country. It can reflect the differences among fresh meat, hypo-fresh meat and degenerative meat according with the sensory changes and is a more objective index. In order to find the corresponding relationship between NH3, H2S, color value, complete adipose cell number per unit area and TVB-N, the paper chose the method of BP neural network as the tool. We took the values of NH3, H2S, H, S, I, complete adipose cell number per unit area as the sample input of the network, took the values of TVB-N as the target output of the network. The arithmetic transferred and the downstage man-machine interface operated made use of communication mechanism between Visual C++ 6.0 and MATLAB. In this way an intellectual identifying system of pork fresh degree was established. Then we put our efforts on the hardware design of the intellectual identifying system. The collection, compression and transmission of the image information as well as the gas sensor data collection and transmission were realized through the portable pork freshness detector on the base of the implementation of DSP technology. The data was then transmitted to the host computer through Ethernet. After image processing and pattern recognition on the host computer, the pork freshness identifying result was then transmitted back to the portable pork freshness detector and displayed on the LCD. The pork freshness detection on the portable terminal was initially realized.In this paper, the intellectual detection method of pork fresh degree based on much data merging was a completely new detection method. It can identify the fresh degree of pork quickly and exactly. The detection result almost accords with the result of TVB-N. It can be used in other related fields and there is high application value about it.