Classification of Bulk Grain Images and Detection of Unwanted Residue in Bulk Grain Images in Neural Networks - Harish Gujjar - Bog

Computers with their inherent ability to store large amount of information have found vast applications in the field of banking, e-commerce, agriculture science, decision support systems, expert systems, image processing etc. In science and Engineering there is a vast scope of applications in Digital image processing in the years to come. There is every possibility of developing a final machine-like human beings where image processing plays an important role. Satellites of remote sensing through spacecrafts, store of images for the purpose of business in banks, processing of images in medicals, manmade machines, acoustic, sonar, radar are some of the list of applications where "Digital Image Processing" DIP is anchored.

In predicting the environment, in searching the resources which are inside the earth, geographical drawings, crop predictions, growth of population, forecasting the weather, etc are carried out by images. "Mars Orbit Mission" MOM, Images are taken in the deep space, in television broadcasting, audio conferencing, video conferencing, duplicating the documents in case of office automation, security systems in banks, wireless military communication are some of the application of images in space. There are such large areas of application where image processing plays a vital role.

In medical applications, X-rays of chest are processed, Cineangio grams, transaxial tomography projection of images, and other images obtained from medical fields that occur in ultrasonic scanning, , nuclear, radiology, magnetic resonance respectively are being carried out. The screening and monitoring of these diseases in patients are done and also detection of trauma is also carried out. Various systems in defense recognizes their targets by using Radar and sonar. There is large listing of applications such as vision of the robot in automation industry, creating a cartoon and in industries of textiles for fashion.

The term "Digital Image processing" (DIP) is processing a 2D image in a digital computer. A large and finite number of bits arranged in a particular order in real or complex scenario are called a digital image. An algorithmic and mathematical tool employed in digital image manipulation is called "Digital Image processing" (DIP).
The process of identifying the bulk grain depending upon their type is called classification and the presence or absence of foreign body is detected are classified as impure or pure. Classification of bulk grain and detection of foreign body in it is the major challenge. This is done by building a appropriate neural network prototype for the classification and recognition of the images of grains with different features with respect to their types and based on foreign body in terms of pure/impure using the feature sets of three alternate sets of shades, appearance, combination of both shades and appearance respectively and detecting of foreign body and type of impure sample.

Digital camera is used to get the images samples of different grains, and image processing techniques are used in separation of features such as color and texture. A classifier is used based on BPNN for the purpose of training. The purity and impurity of the grains is trailed by using the network thus developed. A technique of segmentation and thresholding is used for the impure images to identify the foreign body and category.